History not updated/ui synced

[hexylena]

Describe the bug

I will occasionally start a job and the history will not update to show the outputs. Additionally when I refresh, the history will occasionally not. Today after updating I've been hitting the refresh key far more often than usual.

Galaxy Version and/or server at which you observed the bug

Galaxy Version: 24.1.3.dev0 Commit: 1a6b7fefd373a7a8f8233c08be6bbe30961b6508

Browser and Operating System Operating System: Linux Browser: Chrome Version 130.0.6723.58 (Official Build) (64-bit)

To Reproduce Steps to reproduce the behavior:

1. Run a job (maybe in a large history?)
2. Expand the queued job
3. Click on details
4. Watch the Job Outputs section update and turn green
5. Hover over the history update icon to see it was last refreshed 1 second ago
6. Click the history update button to see it's updated 0 seconds ago, nothing changes, dataset still grey. (Why doesn't this forcibly refresh the history? Rather than just doing the update process that's already happening in the background.)
7. Check network tab to see what's going on. See endlessly requests like /history/current_history_json?since=2024-10-18T11:28:53.259435 that return null

Expected behavior

That when I click refresh history it actually refreshes it.

Screenshots

Additional context

I cannot reproduce this on EU, and I cannot provide access to the server where it happens, but i can run any queries y'all need or apply patches.

[mvdbeek]

Check that the database uses UTC time?

[natefoo]

Possibly #17954? Although in my case the history item text updates, but not the styles.

[hexylena]

Good thought @mvdbeek.

galaxy=# show timezone;
 TimeZone 
----------
 Etc/UTC
(1 row)

[hexylena]

# chronyc -n tracking
Reference ID    : B950F724 (185.80.247.36)
Stratum         : 2
Ref time (UTC)  : Tue Oct 22 09:44:29 2024
System time     : 929.234680176 seconds slow of NTP time

I'm gonna go ahead and guess that might have something to do with it :(

[shenweiyan]

@hexylena @mvdbeek I encountered the same issue in the local Galaxy instance 24.0.4.dev0, it will take a long time for the history to update to display the output.

I found that PRC time is used in my Galaxy database, while my server uses CST time, but it seems that the time difference between them is very short.

Do I need to change the database time to UTC? If so, how should I specifically operate to do that? Thanks.

galaxy=> show timezone;
 TimeZone
----------
 PRC
(1 row)

$ chronyc -n tracking
Reference ID    : 6FE6BDAE (111.230.189.174)
Stratum         : 3
Ref time (UTC)  : Tue Mar 18 07:23:25 2025
System time     : 0.000582530 seconds slow of NTP time