Ondrej Kuznik
01f698bce9
ITS#7256 A few issues discovered during testing slapmodify.
...
- bdb deallocates cursor after some tool operations but expects it to be
around every time it is called
- mdb tries to allocate a second transaction while it already holds one,
deadlocking
- op->o_hdr must exist for most uses of op, but o_tmpmfuncs need not
- what about indexing after a tool modify? Does it happen automatically?
2014-07-08 17:28:12 -07:00
Howard Chu
4afa10d059
ITS#7872 plug memleak of matchedDN
2014-06-04 14:02:06 -07:00
Howard Chu
a01d3f965b
ITS#7870 ignore index DBs in slapcat
2014-06-03 20:38:21 -07:00
Howard Chu
5a08b66141
ITS#7705,#7800 fix prev commit
...
Better check for missing IDs
2014-05-15 21:27:21 -07:00
Howard Chu
437f21b16e
ITS#7705,#7800 fix back-mdb pagedResults search
...
mdb_idscopes was remembering the IDs of every entry it checked;
it should only have been saving the IDs of entries that were
actually in the search scope. The extra entries filled the scopes
array, causing a loop searching for a parent entry that was
never inserted.
2014-05-15 21:06:17 -07:00
Howard Chu
5cab6b77e4
ITS#7798 fix mdb_entry_decode()
2014-02-07 03:02:33 -08:00
Kurt Zeilenga
5c878c1bf2
Happy new year (belated)
2014-01-25 05:21:25 -08:00
Howard Chu
b1ed3bf10a
ITS#7778 fix regression from #7329
2014-01-13 11:26:46 -08:00
Howard Chu
26ce222b6a
ITS#7761 bail out of search if config is pausing
2013-12-10 18:08:36 -08:00
Howard Chu
66bb780b16
ITS#7743 Fix indexed AND filter
...
Partially revert 47f6aec61e
Was using RANGE_FIRST/RANGE_LAST on an IDL known to *not* be a range.
2013-11-11 21:01:33 -08:00
Howard Chu
3d1a0bb17e
ITS#7741 Fix entryDN index lookup
...
DN notfound is authoritative, don't ignore the result
2013-11-06 09:41:42 -08:00
Quanah Gibson-Mount
e70bc5dd64
ITS#7737 Add a matching rule for OlcDbEnvFlgs
2013-11-01 09:57:20 -07:00
Howard Chu
80d8c56afd
ITS#7725 add nordahead envflag
2013-10-12 16:25:33 -07:00
Howard Chu
44504848ac
ITS#7329 optimize index update for simple add ops
...
Don't need to reindex all the attr values if there were no deletes.
2013-10-09 07:55:50 -07:00
Howard Chu
c82dcab9bd
ITS#7720 don't do ptr arithmetic on void *s
2013-10-08 10:40:51 -07:00
Quanah Gibson-Mount
1a712bf18e
Enable features that were hidden behind LDAP_DEVEL
2013-09-19 09:50:52 -07:00
Howard Chu
3b272b43dc
In mdb_checkpoint, force sync
...
Otherwise this is a no-op in various NOSYNC/MAPASYNC situations
2013-09-02 12:54:48 -07:00
Howard Chu
5b2bfec3b4
More for 907b9d4877
...
op.o_extra is only used with threading
2013-08-14 03:51:29 -07:00
Howard Chu
cec27aa517
Fix prev commit
2013-08-12 15:04:51 -07:00
Howard Chu
907b9d4877
Fixes for threaded indexer
...
Works now, but still 2x slower. Leave disabled.
2013-08-12 14:50:54 -07:00
Howard Chu
64f06d5acc
ITS#7662 cleanup reader txn in mdb_entry_get()
2013-08-10 05:05:16 -07:00
Howard Chu
99e22a4366
ITS#7628 fix back-mdb txn refcount
...
If multiple ops used the same opinfo the refcount wasn't being decremented
2013-06-23 19:13:52 -07:00
Howard Chu
a3de7c29ae
ITS#7604 fix modrdn, base-scope search
2013-05-24 11:08:25 -07:00
Howard Chu
3d5e7ef0d2
ITS#7581 more for prev commit
2013-05-23 11:45:05 -07:00
Howard Chu
a8c1438688
ITS#7577 more alias deref checks
2013-05-17 09:15:36 -07:00
Howard Chu
201ddbe3e4
ITS#7577 more for previous commit
2013-04-28 12:47:14 -07:00
Howard Chu
fb537d747c
ITS#7577 alias fixes
2013-04-28 12:25:58 -07:00
Howard Chu
8cc61d8dbd
attrdb fixes
2013-04-05 03:03:07 -07:00
Howard Chu
daf62a0841
ITS#7540 reject undefined attrs
2013-03-19 13:32:42 -07:00
Howard Chu
02b22e7e59
ITS#7542, more for ITS#7432
...
typos in quicksort
2013-03-18 11:43:47 -07:00
Howard Chu
bc1d4424c5
Define MDB_WRITES_PER_COMMIT, default to 500
...
The former default of 1000 may fail for DBs with many indices.
2013-02-23 19:04:13 -08:00
Howard Chu
03b0319b46
ITS#7488 really turn off threaded indexing code
2013-02-19 18:20:17 -08:00
Howard Chu
f827aa683e
ITS#7527 add missing error checks for corrupt index
2013-02-19 18:19:47 -08:00
Howard Chu
9f227f7661
More for dn2id_upgrade
...
handle 0-length suffix
2013-02-10 00:17:51 +00:00
Howard Chu
5c91a27277
Add error checks to prev commit
2013-02-09 19:50:08 +00:00
Howard Chu
2bbee4bb23
Add mdb_dn2id_upgrade
...
Check if upgrade is needed in mdb_db_open()
Run from tool_reindex. Upgrade must not be interrupted.
2013-02-09 19:38:42 +00:00
Howard Chu
33c98df18e
More explicit logging if mdb_txn_commit fails
...
Change LDAP_DEBUG_TRACE to LDAP_DEBUG_ANY.
Add log msg if it wasn't being logged before.
2013-02-02 05:31:39 -08:00
Howard Chu
a8352c03b3
ITS#7473 fix b1a6f6980f
...
uninit'd size in dn2id_walk, bogus pagedresults
2013-01-29 17:15:09 -08:00
Howard Chu
b1a6f6980f
ITS#7473 Scope-based searches
...
Walk subtree if number of subtree entries is smaller than
number of index candidates.
2013-01-26 16:20:49 -08:00
Howard Chu
a650568ef2
ITS#7473 Add mdb_dn2id_walk() to walk subtree
...
Non-recursive tree walker.
2013-01-26 16:19:38 -08:00
Howard Chu
ee8a897ff8
ITS#7473 Setup for subtree entry counts
2013-01-26 16:18:58 -08:00
Howard Chu
2f53035a51
ITS#7496, more for ITS#7483
...
Missing entries need to be checked at runtime too, not just slapcat.
Apparently some people will still try to run slapd even after slapadd
tells them their DB load was invalid.
2013-01-22 12:34:40 +00:00
Howard Chu
fb0b0688b0
ITS#7473 check DN scope before fetching id2e
...
This is only a 10% gain, but still worth having for now.
2013-01-11 03:15:59 -08:00
Howard Chu
acec27ec94
ITS#7488 disable threaded indexer for now
...
No time to find the bug at the moment.
There's no performance benefit anyway.
2013-01-09 09:07:29 -08:00
Kurt Zeilenga
0fd1bf30b8
Happy New Year
2013-01-02 10:22:57 -08:00
Howard Chu
e4ec2be93d
Better handling of missing entries
...
Treat zero-length entries properly (as missing entries) in entry_next()
and entry_get_int().
2012-12-28 11:26:26 -08:00
Howard Chu
d3087229d5
Get MDB txn before acquiring a CSN
...
Since MDB writes are fully serialized in txn_begin, we can use CSNs as
commit sequence numbers, not just change sequence numbers. This will
prevent most instances of out-of-order writes, so syncprov will almost
never need to transmit a change without a cookieCSN. (Out-of-order is
still possible, if the write was received via syncrepl.)
2012-12-13 07:13:39 -08:00
Howard Chu
a2cfb99c52
ITS#7469 fix slapcat
2012-12-10 17:32:44 -08:00
Howard Chu
fe9e1113ae
Rename mdb_open/close mdb_dbi_open/close
2012-12-03 17:06:31 -08:00
Howard Chu
c0f7090cac
Plug slapindex memleak
2012-12-03 11:39:14 -08:00