May 2014 to January 2015
May 2014
Trunk
- Front ported fix for bug CORE-4372 : Deadlock is possible when two data pages contains record fragments pointing to each other
- Front ported: Improvement CORE-4431 : Reduce contention for allocation table lock while database is in stalled physical backup state
- Front ported: Improvement CORE-4432 : Let attachments to not block others when allocation table is read first time
- Front ported: 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.
- Front ported fix for bug CORE-4433 : GlobalRWLock could not downgrade EX lock to SH if readers is present
B2_5_Release
- Improvement CORE-4431 : Reduce contention for allocation table lock while database is in stalled physical backup state
- Improvement CORE-4432 : Let attachments to not block others when allocation table is read first time
- 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.
- Fixed bug CORE-4433 : GlobalRWLock could not downgrade EX lock to SH if readers is present
- Implement support for fallocate() call on Linux.
- Implement improvement CORE-4443 : Use fast file grow on those Linux systems which supports it
- Fixed bug CORE-4444 : Engine could hung and block all attachments in out of disk space condition during physical backup
- Improvement CORE-4445 : Extend main database file faster when physical backup state changed from stalled to merge
B2_1_Release
- Backported fix for bug CORE-2349 : False "Invalid SQLDA" error
June 2014
Trunk
- New boolean setting was added into fbtrace.conf: explain_plan.
- Front ported support for fallocate() call on Linux.
- Front ported improvement CORE-4443 : Use fast file grow on those Linux systems which supports it
- Front ported fix for bug CORE-4444 : Engine could hung and block all attachments in out of disk space condition during physical backup
- Front ported improvement CORE-4445 : Extend main database file faster when physical backup state changed from stalled to merge
- Implement improvement CORE-4451 : Allow output to trace explain plan form
July 2014
Trunk
- Avoid races (and AV as result) when two attachments starts to use temporary page space simultaneously.
- Cleanup and optimize locking a bit (as suggested by Dmitry).While i don't see how we can have positive scale, wrong code should be fixed anyway
August 2014
Trunk
- Fixed bug CORE-4510 : AV in validation is possible
- Missing notification of gc thread.
- Avoid info duplication when statements in call stack attached to different transactions (for example: monitoring snapshot is created in autonomous transaction)
B2_5_Release
- Backport: avoid info duplication when statements in call stack attached to different transactions (for example: monitoring snapshot is created in autonomous transaction)
September 2014
Trunk
- Implement improvement CORE-4556 : Allocate data pages in group of sequential ordered pages (extents)
- Improvement CORE-4562 : Split all validation messages on critical and minor ones
October 2014
Trunk
- Fixed bug CORE-4565 : GDSCODE can have value = 0 in WHEN-section under some concurrent env., only in SC or CS (not in SS)
- Improvement CORE-4562 : Split all validation messages on critical and minor ones.Additional patch, see details in tracker
- Fixed bug CORE-4575 : AV in garbage collector thread at disconnect of last attachment
- Fixed bug CORE-4576 : Cache Writer thread could not start
- Fixed wrong item "unknown ISC error 0" in error messages
- Improved page cache performance a bit.
November 2014
Ttrunk
- Add support for MSVC12 (VS 2013)
- Fixed compilation errors on Windows after IPv6 patches
- Small optimization by Dimitry Sibiryakov:set ppg_min_space to the next really non-full slot when current ppg_min_space slot is marked as full.
- Add check if IPv6 is supported by OS.
- Windows implementation is complete, while POSIX implementation should be reviewed and fixed if necessary.
- Move ConfigRoot from common to the yvalve. Add its methods to the public interface ConfigManager. Now all plug-ins have the same root\install folders.It should fix CORE-4581 and related issues.
- Fix build errors when WIRE_COMPRESS_SUPPORT not defined
B2_5_Release
- Fixed bug CORE-4615 : Classic Server could hung with (near) 100% CPU load
- Fixed bug CORE-4627 : Deadlock with TIP page lock and monitoring lock
December 2014
Trunk
- Add pre-built zlib library
- Support for zlib on Windows
- Patches for mingw by Alexey Pavlov :
- Fix using function pointer
- Mingw compilers don't support Win98
- Mingw also have posix threads library so to resolve conflict we need guard using posix threads for Windows.
- Don't use hack for mingw-w64
January 2015
Trunk
- Fixed my error at patch for CORE-4565. It was the reason of false "lock denied" errors.
- Make shared memory header same size independent on OS.
- Introduce shared memory header version.
- Frontportred fix for bug CORE-2848 : "lock conversion denied" or "lock denied" error
- Improvement CORE-4671 : Release internal temporary blobs early to free some memory\disk space
B2_5_Release
- Improvement CORE-4671 : Release internal temporary blobs early to free some memory\disk space
Vladyslav Khorsun
Dnetpropetrovsk, Ukraine
May 2014 to January 2015
May 2014
Trunk
- Front ported fix for bug CORE-4372 : Deadlock is possible when two data pages contains record fragments pointing to each other
- Front ported: Improvement CORE-4431 : Reduce contention for allocation table lock while database is in stalled physical backup state
- Front ported: Improvement CORE-4432 : Let attachments to not block others when allocation table is read first time
- Front ported: 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.
- Front ported fix for bug CORE-4433 : GlobalRWLock could not downgrade EX lock to SH if readers is present
B2_5_Release
- Improvement CORE-4431 : Reduce contention for allocation table lock while database is in stalled physical backup state
- Improvement CORE-4432 : Let attachments to not block others when allocation table is read first time
- 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.
- Fixed bug CORE-4433 : GlobalRWLock could not downgrade EX lock to SH if readers is present
- Implement support for fallocate() call on Linux.
- Implement improvement CORE-4443 : Use fast file grow on those Linux systems which supports it
- Fixed bug CORE-4444 : Engine could hung and block all attachments in out of disk space condition during physical backup
- Improvement CORE-4445 : Extend main database file faster when physical backup state changed from stalled to merge
B2_1_Release
- Backported fix for bug CORE-2349 : False "Invalid SQLDA" error
June 2014
Trunk
- New boolean setting was added into fbtrace.conf: explain_plan.
- Front ported support for fallocate() call on Linux.
- Front ported improvement CORE-4443 : Use fast file grow on those Linux systems which supports it
- Front ported fix for bug CORE-4444 : Engine could hung and block all attachments in out of disk space condition during physical backup
- Front ported improvement CORE-4445 : Extend main database file faster when physical backup state changed from stalled to merge
- Implement improvement CORE-4451 : Allow output to trace explain plan form
July 2014
Trunk
- Avoid races (and AV as result) when two attachments starts to use temporary page space simultaneously.
- Cleanup and optimize locking a bit (as suggested by Dmitry).While i don't see how we can have positive scale, wrong code should be fixed anyway
August 2014
Trunk
- Fixed bug CORE-4510 : AV in validation is possible
- Missing notification of gc thread.
- Avoid info duplication when statements in call stack attached to different transactions (for example: monitoring snapshot is created in autonomous transaction)
B2_5_Release
- Backport: avoid info duplication when statements in call stack attached to different transactions (for example: monitoring snapshot is created in autonomous transaction)
September 2014
Trunk
- Implement improvement CORE-4556 : Allocate data pages in group of sequential ordered pages (extents)
- Improvement CORE-4562 : Split all validation messages on critical and minor ones
October 2014
Trunk
- Fixed bug CORE-4565 : GDSCODE can have value = 0 in WHEN-section under some concurrent env., only in SC or CS (not in SS)
- Improvement CORE-4562 : Split all validation messages on critical and minor ones.Additional patch, see details in tracker
- Fixed bug CORE-4575 : AV in garbage collector thread at disconnect of last attachment
- Fixed bug CORE-4576 : Cache Writer thread could not start
- Fixed wrong item "unknown ISC error 0" in error messages
- Improved page cache performance a bit.
November 2014
Ttrunk
- Add support for MSVC12 (VS 2013)
- Fixed compilation errors on Windows after IPv6 patches
- Small optimization by Dimitry Sibiryakov:set ppg_min_space to the next really non-full slot when current ppg_min_space slot is marked as full.
- Add check if IPv6 is supported by OS.
- Windows implementation is complete, while POSIX implementation should be reviewed and fixed if necessary.
- Move ConfigRoot from common to the yvalve. Add its methods to the public interface ConfigManager. Now all plug-ins have the same root\install folders.It should fix CORE-4581 and related issues.
- Fix build errors when WIRE_COMPRESS_SUPPORT not defined
B2_5_Release
- Fixed bug CORE-4615 : Classic Server could hung with (near) 100% CPU load
- Fixed bug CORE-4627 : Deadlock with TIP page lock and monitoring lock
December 2014
Trunk
- Add pre-built zlib library
- Support for zlib on Windows
- Patches for mingw by Alexey Pavlov :
- Fix using function pointer
- Mingw compilers don't support Win98
- Mingw also have posix threads library so to resolve conflict we need guard using posix threads for Windows.
- Don't use hack for mingw-w64
January 2015
Trunk
- Fixed my error at patch for CORE-4565. It was the reason of false "lock denied" errors.
- Make shared memory header same size independent on OS.
- Introduce shared memory header version.
- Frontportred fix for bug CORE-2848 : "lock conversion denied" or "lock denied" error
- Improvement CORE-4671 : Release internal temporary blobs early to free some memory\disk space
B2_5_Release
- Improvement CORE-4671 : Release internal temporary blobs early to free some memory\disk space
Vladyslav Khorsun
Dnetpropetrovsk, Ukraine