Zammad-docker-compose: Outgoing e-mails delayed

[veronika]

We installed Zammad via docker-compose. In the last days we had some problems with Zammad: Outgoing e-mails (as well as the e-mails to us saying we got a new ticket) were sent with a large delay (35 hours).

After I noticed that, I upgraded Zammad to the latest version (3.6.0-63). I also upgraded elasticsearch the way it’s explained here: GitHub - zammad/zammad-docker-compose: Zammad Docker images for docker-compose

Afterwards it worked fine. But today it doesn’t. Same problem as before, but with less delay. (Maybe the delay gets larger by the time.)

A few things that might or might not have anything to do with my problem:

1. I still have 50% active shards in elasticsearch container (cluster staus yellow).
2. htop tells me that there are at least 7 scheduler processes running (command: script/scheduler.rb run). They don’t use much CPU but a few of them are running for more than one hour.

My questions:

1. Can I see somewhere in my ticket (or in one of the docker container logs) if the notification e-mail has been sent (to me and all others who are supposed to be informed)?

2. Has anyone had a similar problem or an idea how I can solve this?

• Used Zammad version: 3.6.0-63
• Used Zammad installation source: (source, package, …) zammad-docker-compose
• Operating system: Debian GNU/Linux 9.12 (stretch)
• Browser + version: Chromium 88.0.4324.182
• Docker version: 19.03.8
• Docker-compose version: 1.22.0-rc1

Expected behavior:

• I send an e-mail to our zammad connected e-mail address and expect to get an e-mail that a new ticket has arrived.

Actual behavior:

• The expected e-mail (that tells me that there is a new ticket) is sent with large delay.

Steps to reproduce the behavior:

• I send an e-mail to our zammad connected e-mail address and wait for the e-mail.

[MrGeneration]

Sounds like performance issues.
Email (communication and notification) are background jobs within Zammad.

If e.g. your Elasticsearch reindexing does not work correctly or you absolutely escalated with overviews, you may bring down Zammads processes to it’s knees.

Normally I’d suggest to have a look at our performance tuning section ( Configuration via Environment Variables — Zammad documentation ) but the docker container may not support it. I can’t tell, I’m sorry.

What you should be able to see for sure is ramping up delayed jobs, if you run Delayed::Job.count within a rails console of your container, the number is probably pretty high on your system.