hvlad
0719958901
Fixed bug CORE-5101 : Backup restore is slow in FB3 when the database contains many small tables with indices
2016-02-04 10:33:02 +00:00
alexpeshkoff
08d4a3ed45
Fixed CORE-5044: Pages are encrypted for each shadow separately
2016-02-01 16:16:42 +00:00
dimitr
99ad42ab82
This should fix the races while accessing the relation statistics inside dbb_stats. Reads/writes/fetches/marks are incremented (CCH) and read (INF) unprotected, as it's crash safe and read skews are acceptable. Other counters (both global and relation wise) are protected with a mutex. They're incremented on demand based on the attachment counters (via diffs). While being there, cleaned up the currently unused performance counters.
2016-01-24 21:14:18 +00:00
alexpeshkoff
0891bc723d
Fixed CORE-5041: Database may get corrupted after ALTER DATABASE ENCRYPT/DECRYPT issued multiple times per database open time
2015-12-10 16:58:58 +00:00
dimitr
dc7aad88ed
1) Slightly refactored recheduling and checkout logic.
...
2) Re-enabled checkouts in PIO routines, this resolves CORE-4179. Windows build may be broken, please validate.
3) Implemented faster reaction on asynchronous cancellation/shutdown requests.
4) Avoided locking dbb_sync in AST. This change also solves races during massive attachment shutdown.
2015-11-29 15:12:31 +00:00
roman-simakov
ce8d458e46
Fixed CORE-4978: Improved validation and fix of lost data pages. Thanks to Vlad for help.
2015-11-26 09:28:31 +00:00
hvlad
a6d615c0cd
Fix one of the reason for AV in the engine (see CORE-4680, test by Pavel Zotov at 18 Nov 2015).
...
The race when gone attachment released database object while another attachment is initialized and uses same database object.
Could happens with shared cache (former SS) only.
2015-11-20 10:17:54 +00:00
dimitr
647bad2d61
Fixed attachment lock usage. Minimized chances for races during database shutdown.
2015-11-04 06:06:26 +00:00
dimitr
4e4d8002e5
CORE-4965: Extend the transaction ID space beyond 2^32 transactions. Initially developed in the v2.5 custom branch, then reworked and now merged into v3.0.
2015-10-19 13:32:02 +00:00
alexpeshkoff
139331c7e3
Enhanced memory leaks search - changed FB_NEW to FB_NEW_POOL,
...
plain new to FB_NEW, making allocated with it blocks tagged with file/line.
Make related code slightly better readable.
2015-10-12 14:26:00 +00:00
asfernandes
224025ff7d
Misc.
2015-10-05 01:25:39 +00:00
hvlad
79e9404d50
Fixed bug CORE-4952 : SuperServer could hung on database shutdown
2015-10-03 13:07:17 +00:00
alexpeshkoff
16768b177d
Be more consistent putting IStatus into fast inline wrapper. Also some syntax sugar as suggested by Dmitry
2015-03-27 17:51:19 +00:00
alexpeshkoff
6c5f9b77fe
Cleanup - avoid automatic type conversion in FbLocalStatus as suggested by Claudio
2015-03-23 11:06:51 +00:00
alexpeshkoff
c4c9daf132
Avoid limits on size of status vector in engine. Ensure that status is always clean on entry to interface methods.
2015-03-20 18:02:30 +00:00
robocop
939feab036
Misc.
2015-03-13 15:26:22 +00:00
hvlad
79b760b4ac
Memory leak (CORE-4683)
2015-02-16 13:57:28 +00:00
asfernandes
9ece682387
Misc.
2015-02-14 18:55:00 +00:00
dimitr
11a9b5fc36
Change the status handling a little.
2015-02-11 09:20:51 +00:00
hvlad
a07e2ff547
Frontportred fix for bug CORE-2848 : "lock conversion denied" or "lock denied" error
2015-01-24 11:49:18 +00:00
hvlad
8216554413
Fixed my error at patch for CORE-4565. It was the reason of false "lock denied" errors.
2015-01-21 09:18:18 +00:00
hvlad
f8969d22ca
Improved page cache performance a bit.
...
The case when page buffer is evicted and re-assigned to another page.
2014-10-21 14:53:23 +00:00
alexpeshkoff
1f83c3b1ae
Fixed CORE-4582: Within linger period one can not change some database properties
2014-10-16 11:31:48 +00:00
hvlad
c3a181148a
Fixed bug CORE-4576 : Cache Writer thread could not start
2014-10-10 10:12:23 +00:00
dimitr
7ad1fa43a3
Refactored cache shutdown. This should also fix CORE-4568: Server may crash while disconnecting from the database under load.
2014-10-04 20:10:14 +00:00
hvlad
871830123f
Fixed bug CORE-4565 : GDSCODE can have value = 0 in WHEN-section under some concurrent env., only in SC or CS (not in SS)
2014-10-01 18:17:04 +00:00
alexpeshkoff
f5645faaa0
Fixed windows build
2014-09-30 14:21:44 +00:00
dimitr
b742803ebf
Refactored the shared cache checks. Optimized the shared counter for SS.
2014-09-01 13:19:53 +00:00
dimitr
36af3d8cb2
Cleanup and small refactoring to avoid redundancy.
2014-08-15 14:19:02 +00:00
dimitr
6199136792
Refactored database monitoring, made its handling attachment bound. This should make deadlocks impossible, as well as avoid blocking worker connections while the monitoring request is processed.
2014-07-31 08:56:53 +00:00
alexpeshkoff
532181a7f9
Renamed as suggested by Nickolay and Dmitry
2014-07-28 13:01:10 +00:00
alexpeshkoff
4bdc2a3e19
Replaced macro FB_ALIGN with function fbAlign, minor related cleanup
2014-07-24 06:41:38 +00:00
asfernandes
933dd7daef
Warnings.
2014-07-20 02:19:56 +00:00
asfernandes
86487fda92
Misc.
2014-07-20 02:19:06 +00:00
hvlad
bfb1a6a329
Misc
2014-07-18 12:25:06 +00:00
skidder
e2e901939c
Fix GCC warnings
2014-07-17 21:17:15 +00:00
skidder
7c743fc291
Fix build warnings
2014-07-17 18:48:46 +00:00
hvlad
759bf6a592
Front ported fix for bug CORE-4444 : Engine could hung and block all attachments in out of disk space condition during physical backup
2014-06-10 09:31:35 +00:00
alexpeshkoff
6a2016235a
Fixed CORE-4435: After calling release() for attachment to database (instead detach) in embedded mode attachment remains active forever (and some cleanup re. reference counters)
2014-06-10 07:13:34 +00:00
hvlad
6a806677fd
Front ported:
...
1. Improvement CORE-4431 : Reduce contention for allocation table lock while database is in stalled physical backup state
2. Improvement CORE-4432 : Let attachments to not block others when allocation table is read first time
3. On Windows, file can not be deleted while system writes cached data into it, even if file is not open by anyone.
Therefore flush delta file implicitly before closing it.
2014-05-16 12:07:08 +00:00
robocop
836041e871
Misc.
2014-04-04 07:52:41 +00:00
hvlad
092feed137
Fixed warning by PVS-Studio
2014-02-22 10:01:17 +00:00
hvlad
87ffb96a88
Misc.
2014-01-23 11:20:50 +00:00
robocop
2810026363
Misc and warnings.
2013-12-29 00:35:50 +00:00
dimitr
5453b5c394
Attachments marked as shutdown may still be running (before they reach the rescheduling point), so they should not be accounted as already gone. This should also fix CORE-4236: Database shutdown is reported as successfully completed before all active connections are in fact interrupted.
2013-11-13 15:42:48 +00:00
hvlad
34b99e6f3a
Don't force flush for database which creation is not complete.
2013-11-07 10:32:35 +00:00
dimitr
ddcbc3928c
Unconditionally flush the page cache to disk if FLUSH_ALL / FLUSH_FINI modes are used and FW=OFF. This puts a little bit more consistency and favors explicit flushes over implicit ones (e.g. when database is detached and its file is closed).
2013-09-22 15:35:36 +00:00
asfernandes
a66366bc82
Misc.
2013-08-18 18:53:41 +00:00
dimitr
d2fa09942e
Fixed incorrect accounting re. number of fetches.
2013-08-15 11:54:56 +00:00
hvlad
82b574fbfd
Restore scan-resistance feature of the page cache.
...
Refactor common code a bit.
2013-08-08 08:41:12 +00:00
