This is all I have so far, but I can’t really continue testing without a fix (or at least a work-around) for this one, and the perpetual chown one.
After the upgrade, I cannot upload a file into any library. I tried through the web, through the SeaDrive client, and through the Seafile agent.
The agent log shows these errors:
[12/23/24 15:08:32] http-tx-mgr.c(1160): Transfer repo 'bc309d8f': ('normal', 'init') --> ('normal', 'check')
[12/23/24 15:08:32] http-tx-mgr.c(1160): Transfer repo 'bc309d8f': ('normal', 'check') --> ('normal', 'commit')
[12/23/24 15:08:32] http-tx-mgr.c(2979): Bad response code for PUT https://test-seafile.mydomain.com/seafhttp/repo/bc309d8f-71fb-466d-9558-09b7782aefb5/commit/a0c3126d49c9031d87879338b20ec71f17c8cd43: 500.
[12/23/24 15:08:32] http-tx-mgr.c(3950): Failed to send head commit for repo bc309d8f.
[12/23/24 15:08:32] http-tx-mgr.c(1160): Transfer repo 'bc309d8f': ('normal', 'commit') --> ('error', 'finished')
[12/23/24 15:08:32] sync-mgr.c(640): Repo 'neon' sync state transition from uploading to 'error': 'Server error'.
[12/23/24 15:09:03] sync-mgr.c(602): Repo 'neon' sync state transition from 'synchronized' to 'uploading'.
[12/23/24 15:09:03] http-tx-mgr.c(1160): Transfer repo 'bc309d8f': ('normal', 'init') --> ('normal', 'check')
[12/23/24 15:09:04] http-tx-mgr.c(1160): Transfer repo 'bc309d8f': ('normal', 'check') --> ('normal', 'commit')
[12/23/24 15:09:04] http-tx-mgr.c(2979): Bad response code for PUT https://test-seafile.mydomain.com/seafhttp/repo/bc309d8f-71fb-466d-9558-09b7782aefb5/commit/a0c3126d49c9031d87879338b20ec71f17c8cd43: 500.
[12/23/24 15:09:04] http-tx-mgr.c(3950): Failed to send head commit for repo bc309d8f.
[12/23/24 15:09:04] http-tx-mgr.c(1160): Transfer repo 'bc309d8f': ('normal', 'commit') --> ('error', 'finished')
[12/23/24 15:09:04] sync-mgr.c(640): Repo 'neon' sync state transition from uploading to 'error': 'Server error'.
And the server log says:
[2024-12-23 15:08:32] [ERROR] path /repo/bc309d8f-71fb-466d-9558-09b7782aefb5/commit/a0c3126d49c9031d87879338b20ec71f17c8cd43 internal server error: Failed to save gc id: Error 1146: Table 'seafile_db.GCID' doesn't exist
It’s strange because the upgrade script did create other tables, like OrgDownloadRateLimit and OrgUploadRateLimit.
Running a seaf-gc didn’t create that table, but it did complain many times about the absence of that table. It also claimed that every library is broken and should be fixed with seaf-fsck, but the seaf-fsck reported no problems with any library. Also the seaf-fsck did not fix the problem.