curl/tests/http/scorecard.py

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

640 lines
25 KiB
Python
Raw Normal View History

connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
#***************************************************************************
# _ _ ____ _
# Project ___| | | | _ \| |
# / __| | | | |_) | |
# | (__| |_| | _ <| |___
# \___|\___/|_| \_\_____|
#
# Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
#
# This software is licensed as described in the file COPYING, which
# you should have received as part of this distribution. The terms
# are also available at https://curl.se/docs/copyright.html.
#
# You may opt to use, copy, modify, merge, publish, distribute and/or sell
# copies of the Software, and permit persons to whom the Software is
# furnished to do so, under the terms of the COPYING file.
#
# This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
# KIND, either express or implied.
#
# SPDX-License-Identifier: curl
#
###########################################################################
#
import argparse
import json
import logging
import os
import re
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
import sys
from statistics import mean
from typing import Dict, Any, Optional, List
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
from testenv import Env, Httpd, Nghttpx, CurlClient, Caddy, ExecResult, NghttpxQuic, RunProfile
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
log = logging.getLogger(__name__)
class ScoreCardException(Exception):
pass
class ScoreCard:
def __init__(self, env: Env,
httpd: Optional[Httpd],
nghttpx: Optional[Nghttpx],
caddy: Optional[Caddy],
verbose: int,
curl_verbose: int):
self.verbose = verbose
self.env = env
self.httpd = httpd
self.nghttpx = nghttpx
self.caddy = caddy
self._silent_curl = not curl_verbose
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
def info(self, msg):
if self.verbose > 0:
sys.stderr.write(msg)
sys.stderr.flush()
def handshakes(self, proto: str) -> Dict[str, Any]:
props = {}
sample_size = 5
self.info(f'TLS Handshake\n')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
for authority in [
'curl.se', 'google.com', 'cloudflare.com', 'nghttp2.org'
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
]:
self.info(f' {authority}...')
props[authority] = {}
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
for ipv in ['ipv4', 'ipv6']:
self.info(f'{ipv}...')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
c_samples = []
hs_samples = []
errors = []
for i in range(sample_size):
curl = CurlClient(env=self.env, silent=self._silent_curl)
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
args = [
'--http3-only' if proto == 'h3' else '--http2',
f'--{ipv}', f'https://{authority}/'
]
r = curl.run_direct(args=args, with_stats=True)
if r.exit_code == 0 and len(r.stats) == 1:
c_samples.append(r.stats[0]['time_connect'])
hs_samples.append(r.stats[0]['time_appconnect'])
else:
errors.append(f'exit={r.exit_code}')
props[authority][f'{ipv}-connect'] = mean(c_samples) \
if len(c_samples) else -1
props[authority][f'{ipv}-handshake'] = mean(hs_samples) \
if len(hs_samples) else -1
props[authority][f'{ipv}-errors'] = errors
self.info('ok.\n')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
return props
def _make_docs_file(self, docs_dir: str, fname: str, fsize: int):
fpath = os.path.join(docs_dir, fname)
data1k = 1024*'x'
flen = 0
with open(fpath, 'w') as fd:
while flen < fsize:
fd.write(data1k)
flen += len(data1k)
return flen
def _check_downloads(self, r: ExecResult, count: int):
error = ''
if r.exit_code != 0:
error += f'exit={r.exit_code} '
if r.exit_code != 0 or len(r.stats) != count:
error += f'stats={len(r.stats)}/{count} '
fails = [s for s in r.stats if s['response_code'] != 200]
if len(fails) > 0:
error += f'{len(fails)} failed'
return error if len(error) > 0 else None
def transfer_single(self, url: str, proto: str, count: int):
sample_size = count
count = 1
samples = []
errors = []
profiles = []
self.info(f'single...')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
for i in range(sample_size):
curl = CurlClient(env=self.env, silent=self._silent_curl)
r = curl.http_download(urls=[url], alpn_proto=proto, no_save=True,
with_headers=False, with_profile=True)
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
err = self._check_downloads(r, count)
if err:
errors.append(err)
else:
total_size = sum([s['size_download'] for s in r.stats])
samples.append(total_size / r.duration.total_seconds())
profiles.append(r.profile)
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
return {
'count': count,
'samples': sample_size,
'speed': mean(samples) if len(samples) else -1,
'errors': errors,
'stats': RunProfile.AverageStats(profiles),
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
}
def transfer_serial(self, url: str, proto: str, count: int):
sample_size = 1
samples = []
errors = []
profiles = []
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
url = f'{url}?[0-{count - 1}]'
self.info(f'serial...')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
for i in range(sample_size):
curl = CurlClient(env=self.env, silent=self._silent_curl)
r = curl.http_download(urls=[url], alpn_proto=proto, no_save=True,
with_headers=False, with_profile=True)
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
err = self._check_downloads(r, count)
if err:
errors.append(err)
else:
total_size = sum([s['size_download'] for s in r.stats])
samples.append(total_size / r.duration.total_seconds())
profiles.append(r.profile)
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
return {
'count': count,
'samples': sample_size,
'speed': mean(samples) if len(samples) else -1,
'errors': errors,
'stats': RunProfile.AverageStats(profiles),
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
}
def transfer_parallel(self, url: str, proto: str, count: int):
sample_size = 1
samples = []
errors = []
profiles = []
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
url = f'{url}?[0-{count - 1}]'
self.info(f'parallel...')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
for i in range(sample_size):
curl = CurlClient(env=self.env, silent=self._silent_curl)
r = curl.http_download(urls=[url], alpn_proto=proto, no_save=True,
with_headers=False,
with_profile=True,
extra_args=['--parallel',
'--parallel-max', str(count)])
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
err = self._check_downloads(r, count)
if err:
errors.append(err)
else:
total_size = sum([s['size_download'] for s in r.stats])
samples.append(total_size / r.duration.total_seconds())
profiles.append(r.profile)
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
return {
'count': count,
'samples': sample_size,
'speed': mean(samples) if len(samples) else -1,
'errors': errors,
'stats': RunProfile.AverageStats(profiles),
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
}
def download_url(self, label: str, url: str, proto: str, count: int):
self.info(f' {count}x{label}: ')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
props = {
'single': self.transfer_single(url=url, proto=proto, count=10),
}
if count > 1:
props['serial'] = self.transfer_serial(url=url, proto=proto,
count=count)
props['parallel'] = self.transfer_parallel(url=url, proto=proto,
count=count)
self.info(f'ok.\n')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
return props
def downloads(self, proto: str, count: int,
fsizes: List[int]) -> Dict[str, Any]:
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
scores = {}
if self.httpd:
if proto == 'h3':
port = self.env.h3_port
via = 'nghttpx'
descr = f'port {port}, proxying httpd'
else:
port = self.env.https_port
via = 'httpd'
descr = f'port {port}'
self.info(f'{via} downloads\n')
scores[via] = {
'description': descr,
}
for fsize in fsizes:
label = self.fmt_size(fsize)
fname = f'score{label}.data'
self._make_docs_file(docs_dir=self.httpd.docs_dir,
fname=fname, fsize=fsize)
url = f'https://{self.env.domain1}:{port}/{fname}'
results = self.download_url(label=label, url=url,
proto=proto, count=count)
scores[via][label] = results
if self.caddy:
port = self.caddy.port
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
via = 'caddy'
descr = f'port {port}'
self.info('caddy downloads\n')
scores[via] = {
'description': descr,
}
for fsize in fsizes:
label = self.fmt_size(fsize)
fname = f'score{label}.data'
self._make_docs_file(docs_dir=self.caddy.docs_dir,
fname=fname, fsize=fsize)
url = f'https://{self.env.domain1}:{port}/{fname}'
results = self.download_url(label=label, url=url,
proto=proto, count=count)
scores[via][label] = results
return scores
def do_requests(self, url: str, proto: str, count: int,
max_parallel: int = 1):
sample_size = 1
samples = []
errors = []
profiles = []
url = f'{url}?[0-{count - 1}]'
extra_args = ['--parallel', '--parallel-max', str(max_parallel)] \
if max_parallel > 1 else []
self.info(f'{max_parallel}...')
for i in range(sample_size):
curl = CurlClient(env=self.env, silent=self._silent_curl)
r = curl.http_download(urls=[url], alpn_proto=proto, no_save=True,
with_headers=False, with_profile=True,
extra_args=extra_args)
err = self._check_downloads(r, count)
if err:
errors.append(err)
else:
for _ in r.stats:
samples.append(count / r.duration.total_seconds())
profiles.append(r.profile)
return {
'count': count,
'samples': sample_size,
'speed': mean(samples) if len(samples) else -1,
'errors': errors,
'stats': RunProfile.AverageStats(profiles),
}
def requests_url(self, url: str, proto: str, count: int):
self.info(f' {url}: ')
props = {
'1': self.do_requests(url=url, proto=proto, count=count),
'6': self.do_requests(url=url, proto=proto, count=count,
max_parallel=6),
'25': self.do_requests(url=url, proto=proto, count=count,
max_parallel=25),
'50': self.do_requests(url=url, proto=proto, count=count,
max_parallel=50),
'100': self.do_requests(url=url, proto=proto, count=count,
max_parallel=100),
}
self.info(f'ok.\n')
return props
def requests(self, proto: str, req_count) -> Dict[str, Any]:
scores = {}
if self.httpd:
if proto == 'h3':
port = self.env.h3_port
via = 'nghttpx'
descr = f'port {port}, proxying httpd'
else:
port = self.env.https_port
via = 'httpd'
descr = f'port {port}'
self.info(f'{via} requests\n')
self._make_docs_file(docs_dir=self.httpd.docs_dir,
fname='reqs10.data', fsize=10*1024)
url1 = f'https://{self.env.domain1}:{port}/reqs10.data'
scores[via] = {
'description': descr,
'count': req_count,
'10KB': self.requests_url(url=url1, proto=proto, count=req_count),
}
if self.caddy:
port = self.caddy.port
via = 'caddy'
descr = f'port {port}'
self.info('caddy requests\n')
self._make_docs_file(docs_dir=self.caddy.docs_dir,
fname='req10.data', fsize=10 * 1024)
url1 = f'https://{self.env.domain1}:{port}/req10.data'
scores[via] = {
'description': descr,
'count': req_count,
'10KB': self.requests_url(url=url1, proto=proto, count=req_count),
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
}
return scores
def score_proto(self, proto: str,
handshakes: bool = True,
downloads: Optional[List[int]] = None,
download_count: int = 50,
req_count=5000,
requests: bool = True):
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
self.info(f"scoring {proto}\n")
p = {}
if proto == 'h3':
p['name'] = 'h3'
if not self.env.have_h3_curl():
raise ScoreCardException('curl does not support HTTP/3')
for lib in ['ngtcp2', 'quiche', 'msh3', 'nghttp3']:
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
if self.env.curl_uses_lib(lib):
p['implementation'] = lib
break
elif proto == 'h2':
p['name'] = 'h2'
if not self.env.have_h2_curl():
raise ScoreCardException('curl does not support HTTP/2')
for lib in ['nghttp2', 'hyper']:
if self.env.curl_uses_lib(lib):
p['implementation'] = lib
break
elif proto == 'h1' or proto == 'http/1.1':
proto = 'http/1.1'
p['name'] = proto
p['implementation'] = 'hyper' if self.env.curl_uses_lib('hyper')\
else 'native'
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
else:
raise ScoreCardException(f"unknown protocol: {proto}")
if 'implementation' not in p:
raise ScoreCardException(f'did not recognized {p} lib')
p['version'] = Env.curl_lib_version(p['implementation'])
score = {
'curl': self.env.curl_fullname(),
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
'os': self.env.curl_os(),
'protocol': p,
}
if handshakes:
score['handshakes'] = self.handshakes(proto=proto)
if downloads and len(downloads) > 0:
score['downloads'] = self.downloads(proto=proto,
count=download_count,
fsizes=downloads)
if requests:
score['requests'] = self.requests(proto=proto, req_count=req_count)
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
self.info("\n")
return score
def fmt_ms(self, tval):
return f'{int(tval*1000)} ms' if tval >= 0 else '--'
def fmt_size(self, val):
if val >= (1024*1024*1024):
return f'{val / (1024*1024*1024):0.000f}GB'
elif val >= (1024 * 1024):
return f'{val / (1024*1024):0.000f}MB'
elif val >= 1024:
return f'{val / 1024:0.000f}KB'
else:
return f'{val:0.000f}B'
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
def fmt_mbs(self, val):
return f'{val/(1024*1024):0.000f} MB/s' if val >= 0 else '--'
def fmt_reqs(self, val):
return f'{val:0.000f} r/s' if val >= 0 else '--'
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
def print_score(self, score):
print(f'{score["protocol"]["name"].upper()} in {score["curl"]}')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
if 'handshakes' in score:
print(f'{"Handshakes":<24} {"ipv4":25} {"ipv6":28}')
print(f' {"Host":<17} {"Connect":>12} {"Handshake":>12} '
f'{"Connect":>12} {"Handshake":>12} {"Errors":<20}')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
for key, val in score["handshakes"].items():
print(f' {key:<17} {self.fmt_ms(val["ipv4-connect"]):>12} '
f'{self.fmt_ms(val["ipv4-handshake"]):>12} '
f'{self.fmt_ms(val["ipv6-connect"]):>12} '
f'{self.fmt_ms(val["ipv6-handshake"]):>12} '
f'{"/".join(val["ipv4-errors"] + val["ipv6-errors"]):<20}'
)
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
if 'downloads' in score:
# get the key names of all sizes and measurements made
sizes = []
measures = []
m_names = {}
mcol_width = 12
mcol_sw = 17
for server, server_score in score['downloads'].items():
for sskey, ssval in server_score.items():
if isinstance(ssval, str):
continue
if sskey not in sizes:
sizes.append(sskey)
for mkey, mval in server_score[sskey].items():
if mkey not in measures:
measures.append(mkey)
m_names[mkey] = f'{mkey}({mval["count"]}x)'
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
print('Downloads')
print(f' {"Server":<8} {"Size":>8}', end='')
for m in measures: print(f' {m_names[m]:>{mcol_width}} {"[cpu/rss]":<{mcol_sw}}', end='')
print(f' {"Errors":^20}')
for server in score['downloads']:
for size in sizes:
size_score = score['downloads'][server][size]
print(f' {server:<8} {size:>8}', end='')
errors = []
for key, val in size_score.items():
if 'errors' in val:
errors.extend(val['errors'])
for m in measures:
if m in size_score:
print(f' {self.fmt_mbs(size_score[m]["speed"]):>{mcol_width}}', end='')
s = f'[{size_score[m]["stats"]["cpu"]:>.1f}%'\
f'/{self.fmt_size(size_score[m]["stats"]["rss"])}]'
print(f' {s:<{mcol_sw}}', end='')
else:
print(' '*mcol_width, end='')
if len(errors):
print(f' {"/".join(errors):<20}')
else:
print(f' {"-":^20}')
if 'requests' in score:
sizes = []
measures = []
m_names = {}
mcol_width = 9
mcol_sw = 13
for server in score['requests']:
server_score = score['requests'][server]
for sskey, ssval in server_score.items():
if isinstance(ssval, str) or isinstance(ssval, int):
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
continue
if sskey not in sizes:
sizes.append(sskey)
for mkey, mval in server_score[sskey].items():
if mkey not in measures:
measures.append(mkey)
m_names[mkey] = f'{mkey}'
print('Requests, max in parallel')
print(f' {"Server":<8} {"Size":>6} {"Reqs":>6}', end='')
for m in measures: print(f' {m_names[m]:>{mcol_width}} {"[cpu/rss]":<{mcol_sw}}', end='')
print(f' {"Errors":^10}')
for server in score['requests']:
for size in sizes:
size_score = score['requests'][server][size]
count = score['requests'][server]['count']
print(f' {server:<8} {size:>6} {count:>6}', end='')
errors = []
for key, val in size_score.items():
if 'errors' in val:
errors.extend(val['errors'])
for m in measures:
if m in size_score:
print(f' {self.fmt_reqs(size_score[m]["speed"]):>{mcol_width}}', end='')
s = f'[{size_score[m]["stats"]["cpu"]:>.1f}%'\
f'/{self.fmt_size(size_score[m]["stats"]["rss"])}]'
print(f' {s:<{mcol_sw}}', end='')
else:
print(' '*mcol_width, end='')
if len(errors):
print(f' {"/".join(errors):<10}')
else:
print(f' {"-":^10}')
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
def parse_size(s):
m = re.match(r'(\d+)(mb|kb|gb)?', s, re.IGNORECASE)
if m is None:
raise Exception(f'unrecognized size: {s}')
size = int(m.group(1))
if not m.group(2):
pass
elif m.group(2).lower() == 'kb':
size *= 1024
elif m.group(2).lower() == 'mb':
size *= 1024 * 1024
elif m.group(2).lower() == 'gb':
size *= 1024 * 1024 * 1024
return size
def main():
parser = argparse.ArgumentParser(prog='scorecard', description="""
Run a range of tests to give a scorecard for a HTTP protocol
'h3' or 'h2' implementation in curl.
""")
parser.add_argument("-v", "--verbose", action='count', default=1,
help="log more output on stderr")
parser.add_argument("-j", "--json", action='store_true',
default=False, help="print json instead of text")
parser.add_argument("-H", "--handshakes", action='store_true',
default=False, help="evaluate handshakes only")
parser.add_argument("-d", "--downloads", action='store_true',
default=False, help="evaluate downloads")
parser.add_argument("--download", action='append', type=str,
default=None, help="evaluate download size")
parser.add_argument("--download-count", action='store', type=int,
default=50, help="perform that many downloads")
parser.add_argument("-r", "--requests", action='store_true',
default=False, help="evaluate requests")
parser.add_argument("--request-count", action='store', type=int,
default=5000, help="perform that many requests")
parser.add_argument("--httpd", action='store_true', default=False,
help="evaluate httpd server only")
parser.add_argument("--caddy", action='store_true', default=False,
help="evaluate caddy server only")
parser.add_argument("--curl-verbose", action='store_true',
default=False, help="run curl with `-v`")
parser.add_argument("protocol", default='h2', nargs='?',
help="Name of protocol to score")
args = parser.parse_args()
if args.verbose > 0:
console = logging.StreamHandler()
console.setLevel(logging.INFO)
console.setFormatter(logging.Formatter(logging.BASIC_FORMAT))
logging.getLogger('').addHandler(console)
protocol = args.protocol
handshakes = True
downloads = [1024 * 1024, 10 * 1024 * 1024, 100 * 1024 * 1024]
if args.download is not None:
downloads = []
for x in args.download:
downloads.extend([parse_size(s) for s in x.split(',')])
requests = True
if args.downloads or args.requests or args.handshakes:
handshakes = args.handshakes
if not args.downloads:
downloads = None
requests = args.requests
test_httpd = protocol != 'h3'
test_caddy = True
if args.caddy or args.httpd:
test_caddy = args.caddy
test_httpd = args.httpd
rv = 0
env = Env()
env.setup()
env.test_timeout = None
httpd = None
nghttpx = None
caddy = None
try:
if test_httpd:
httpd = Httpd(env=env)
assert httpd.exists(), \
f'httpd not found: {env.httpd}'
httpd.clear_logs()
assert httpd.start()
if 'h3' == protocol:
nghttpx = NghttpxQuic(env=env)
nghttpx.clear_logs()
assert nghttpx.start()
if test_caddy and env.caddy:
caddy = Caddy(env=env)
caddy.clear_logs()
assert caddy.start()
card = ScoreCard(env=env, httpd=httpd, nghttpx=nghttpx, caddy=caddy,
verbose=args.verbose, curl_verbose=args.curl_verbose)
score = card.score_proto(proto=protocol,
handshakes=handshakes,
downloads=downloads,
download_count=args.download_count,
req_count=args.request_count,
requests=requests)
if args.json:
print(json.JSONEncoder(indent=2).encode(score))
else:
card.print_score(score)
except ScoreCardException as ex:
sys.stderr.write(f"ERROR: {str(ex)}\n")
rv = 1
except KeyboardInterrupt:
log.warning("aborted")
rv = 1
finally:
if caddy:
caddy.stop()
if nghttpx:
nghttpx.stop(wait_dead=False)
if httpd:
httpd.stop()
sys.exit(rv)
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
if __name__ == "__main__":
main()