October 2019 to January 2020
January 2020
- Created / modified QA tests for > 30 tracker tickets
- Found / reported bugs: CORE-6218, CORE-6223, CORE-6227, CORE-6238
- Reported inconsistencies in FB 4.0 (Srp vs Legacy) while refactored old test for CORE-4766
- Reported ACTION_TRACE_STOP not forcing the trace log to be fully written (flushed) to disk
- Reported significant time degradation in test runs for Srp as compared to LegacyAuth
- Found bug in FDB driver: Transaction::execute_immediate() method does not send the SQL dialect properly
- Reported the fix for CORE-6236 being incomplete
- Found regressions in fixes for CORE-4135 and CORE-5974
- Multiple improvements in the fbt-cross-report package
December 2019
- Created / modified QA tests for >30 tracker tickets
- Adjusted serveral old tests related to optimizer by Dmitry Yemanov's request
- Found / reported bugs: CORE-6204, CORE-6211
- fbt-cross-report: continued implementing
- Reported FB crash caused by 'gfix -shut full -force 0 localhost:'
- Refactored rundaily.bat: check for FB crashes and create backtraces in this case
November 2019
- Created / modified QA tests for 10 tracker tickets
- Found / reported bugs: CORE-6182, CORE-6183, CORE-6185, CORE-6186, CORE-6187, CORE-6196
- Reported SC hanging during the database shutdown under high load
- Compared performance for database in normal vs locked (by nbackup) modes
- Checked cursor behavior (stable vs unstable) when used in positioned updates
- Reported "gstat -r" causing "Unexpected end of DB file" when DB is in backup-lock state
- Reported several queries with FULL SCANs causing the time of backup to significantly increase
- Reported the engine continuously trying to attach to a database in the full shutdown mode
- fbt-cross-report: initial development
October 2019
- Created / modified QA tests for 18 tracker tickets
- Re-checked known SIMILAR TO problems against the new 're2' library implementation
- Reported problem with test for CORE-6097 when system locale differs from 419 (russian)
- Error encountered when creating a sequence starting with value greater than 2^31-1 (probably FDB driver bug)
Pavel Zotov
Moscow, Russian Federation
