What is your original issue/pain point you want to solve?
Container restarts take ~5 minutes because Translation.sync downloads
translations every time.
Which are one or two concrete situations where this problem hurts the most?
After changing settings, we have to wait 5+ minutes for restart
Many “init container not completing” reports (#101, #79, #180, #384)
may be caused by users not waiting long enough
Why is it not solvable with the Zammad standard?
Translation.sync is hardcoded in docker-entrypoint.sh with no skip option.
What is your expectation/what do you want to achieve?
Add environment variable to skip translation sync.
This would reduce restart time from ~5 min to ~2 min.
Users can manually sync via Admin UI when needed.
Your Zammad environment:
Average concurrent agent count: Small team
Average tickets a day: N/A (new installation)
What roles/people are involved: System administrator
Sorry, but no. This is not how it works since a very long time.
Translations are synced from Zammads storage (they’re delivered via source files).
The translation sync -is required- after updates and should stay enforced.
However, what I do understand is that the stack should have some way to tell if it actually is updating or just restarting. Hard to accomplish I guess.
I now understand that the translation sync runs from source files rather than downloading each time. That makes sense.
I agree that syncing after updates is essential and should continue.
I appreciate your point about distinguishing between updates and simple restarts being ideal.
Translation.sync does run every time, but at a point where the rest of the start is not blocked any longer. Other containers like the Railsserver only wait for the migrations to be executed, and then they start, while zammad-init is still running. So it does take a bit of time, but it does not block stack restarts. If that were the case, then we should need to fix it.