Buildroots of passing builds are taking far too long to be cleaned up, despite a delay of only 2 minutes being defined in koji/daemon.py [1]. Instead, the logs show that cleanup only occurs after approximately 8h2m. The extra 8h coincides with the number of hours by which my timezone (Asia/Kuala_Lumpur) is ahead of GMT.
From koji call getBuildroot:
koji call getBuildroot
{ ... 'retire_event_id': 52964, 'retire_event_time': '2019-07-27 18:45:36.655608', 'retire_ts': 1564253136.65561, ... }
retire_event_time is correct when treated as local time, and is consistent with that displayed in the web UI. However, date -d @1564253136.65561 returns Sun 28 Jul 2019 02:45:36 AM +08, 8h ahead of retire_event_time.
retire_event_time
date -d @1564253136.65561
Sun 28 Jul 2019 02:45:36 AM +08
kojid calulates buildroot age as time.time() - br['retire_ts'], where time() is the epoch in UTC [2]. Due to the above quirk in retire_ts, it takes 8h before kojid catches up to a retired buildroot.
time.time() - br['retire_ts']
time()
retire_ts
koji 1.15.2 postgresql 9.6.13
Metadata Update from @tkopecek: - Custom field Size adjusted to None - Issue tagged with: bug
This should have been fixed by #2160
Metadata Update from @mikem: - Issue close_status updated to: Fixed - Issue status updated to: Closed (was: Open)
This issue has been migrated to Fedora Forge: https://forge.fedoraproject.org/koji/koji/issues/1586
Please continue any further discussion there.