Elastic search always out of sync and no schedule

Technical assistance
1

Infos:

  • Used Zammad version: 7.0.2-1781669454.9dc15acf.centos9
  • Used Zammad installation type: (source, package, docker-compose, …) package
  • Operating system: Oracle Linux 9
  • Browser + version: Edge 149.0.4022.98

Expected behavior:

  • All records are indexed and available on the app search - tickets, users, and articles.
  • Scheduled tasks run as normal

Actual behavior:

  • Something seems broken on my zammad install, as I have to reindex ElasticSearch everyday using zammad run rake zammad:searchindex:rebuild[12].
  • Right after the reindex, all search functionality from the app works, but it starts to fail to find new tickets and users a few hours later.
  • My scheduled tasks (reminders, and triggers) do not start when they are supposed to.

Steps to reproduce the behavior:

  • Package install on RHEL
  • Connection to ElasticSearch 8.19.4 on same host
  • Change config files to connect to another postgres cluster off host

Any advice on what i could check to make sync and scheduler work as expected?
Thank you!

2

Have you checked Zammads monitoring endpoint? I’d guess that your background worker is at least partially not running and you have stacking background jobs. Hard to tell with the provided information, could also be a performance issue but well.

3

Hi, MrGeneration.

I was not aware of that Monitoring page. Nothing is reported there and the instance is healthy
{
“healthy”: true,
“message”: “success”,
“issues”: ,
“actions”: ,
“token”: “123”
}

Logs also seem to show that things are working but I already had to reindex elasticsearch twice today - zammad-worker is also running on systemctl.
Any other process I can check to make sure its actually healthy?

Thanks.

4

How do you qualify that the searchindex gets out of sync…?

5

Everything that is new - customers and tickets, for example, from the day are never returned on the search engine on the interface. Once I run the reindex, they start to show up. It seems to be working for the next hour or so then the new records are not found again.

6

That doesn‘t seem right. Any performance tunings set?

7

No, nothing special. This was working ok while the app was installed on a Debian-package version. Because of changes on cloud providers, once we moved to Oracle Cloud (and to Oracle Linux VM) this problem started. After this change, I had to resync every 15 days, and now every day.

This morning I have deleted one Microsoft Exchange integration from the app and that seems to have improved the overall performance, with everything working as expected.
Would that be a bottleneck and cause this weird behaviour?