January 2023

• Implemented / refactored /adjusted 10 tests from FBT suite.
• Lot of changes in QA-rundaily .sh / .bat scenarios after regular jobs started.
• Fully re-implemented Python script that makes final HTML report to uploading to HTTP server.
• NEW QA. LI-5.0.0.884: hangs on blob_not_found_in_rw_replica_if_target_row_exists.py test detected.NEW QA: several crashes when running tests on Linux (CentOS-7), builds: 5.0.0.863, .865, .877, .884, .888; 4.0.3.2885.
• NEW QA. LI-5.0.0.865, CS: hangs on test read_consist_sttm_restart_on_update_04.py.
• NEW QA. LINUX, 4.0.3.2878 SS: crash on replication-related "test shutdown during applying segments leads to crash.py".
• NEW QA. It will be useful to see in cross-report details info about crashes occurred for last 30 runs.
• NEW QA. Detected hang at final point of tests session when ran pytest on Linux, 4.0.3.2885, experimental build. May relate to #7359.
• NEW QA. Questions about how trace API is used in the firebird-qa. Discussed with pcisar et al, letters 09.01.23 …18.01.23
• Discussed way to obtain failed thread details from stack-trace on Windows (in order to put this info in detalization page for error outcome of some test if it caused by FB crash).
• Attributes 'no-reserve' and default sweep (20000) are 'silently' added to DB header after b/r if source DB was in read-only mode. Created ticket gh-7453.
• NEW QA. LI-5.0.0.907, CS, crashed when performing functional.replication.test_ddl_triggers_must_not_fire_on_replica
• Disabled functional/basic/db_01 ... db_31 in fbtest and in new qa framework after discussion with dimitr.
• Discussed with dimitr needed corrections in the test for core-5061 (plan length exceeded 64K after adding details in recent FB snapshots).
• Test shutdown_during_applying_segments_leads_to_crash.py caused crash with weird message in gdb: "PC register is not available".
• Updated QA plugin to version 0.17.2, checked again set of tests on Linux which caused pytest to hang (FB 4.x).
• NEW QA. Started regular runs of cron jobs on Linux CentOS-7, with uploading final reports on http server (FB 5.x, 4.x and 3.x).
• Miscelaneous corrections of batch scenario.
• Failed to create snapshot WI-4.0.3.2885 (experimental build) with "keyholder.dll : fatal error LNK1120: 2 unresolved externals".
• Two ISQL processes crashed during perform test for CORE-5275 (LI-5.0.0.916, SS), appear only in NEW QA.
• NEW QA, script for creating html report: misc improvements related to interface (tooltips etc)
• NEW QA, LI-5.0.0.917 SS: hanging test read_consist_sttm_restart_max_limit (no such problem in old fbtest suite).
• Linux Debian-10: upgraded Python to 3.10 in order to create virtual env for starting NEW QA together with old FBT (that uses Python-2.7), decided that they will work together for several months.
• Problem with build LI-4.0.3.2891 on CentOS-7 ("cannot create /chroot/.../gen/Firebird--.tar.xz: Directory nonexistent").
• OLTP-EMUL, LI-3.0.11.33657 SS: crash after 15 minutes of work.
• Collation based on ICU prevents DB migration to another host with another ICU version. Created ticket #7463
• NEW QA and olf fbtest jobs, working together since 27.01.2023. Schedule must be changed in order to have all results during ~12…14 hours. Job 'increment build number' must be done about midnight because of this.
• Snapshot 4.0.3.2891 (experimental build) could not be created if standard FB 4.x is running at this time: problem with access to shmem file ("Wrong file for memory mapping, see details in firebird.log").
• Tested special build with fix of problem when record length exceeds 64K but this limit is pretty far from sum of formats:
  "gbak: ERROR:message length error (encountered 32, expected 65568)", requested by Alex  ('wideburp').
• OLTP-EMUL, LI-V4.0.3.2890 SS: crash after 15 minutes of work.