BUG/Enhancement - Elastic-search init container

[RegisHubelia]

Bug/Enhancement request

• Currently, when a zammad container start, elasticsearch-init can take a long time to index if you have a lot of data.
• On top of that, if you want 2 replicas of zammad, the second replica also re-index all the data.
• it then takes quite some time when upgrading/changing configuration for the app to be available. For us it's about 20 minutes per pod - and we have 2 - so about 40 minutes until app/search works as expected.

What you expected to happen:

• The init container should check if elastic search indexes is already initialised and contains data. If so, ignore re-indexing.
• Using redis pub/sub to add/remove items from the indexes - my guess is that it's probably already the case as new tickets gets added in the indexes.
• Scheduled full re-index on a specified/configured cron schedule?

This way, indexing wouldn't block the app availability, and we'd still be quite comfortable with the accuracy of the indexes.

[MrGeneration]

This is not a bug but expected behavior.
You may want to check the configuration options you have and adjust your configuration accordingly.

https://github.com/zammad/zammad-helm/blob/main/zammad/values.yaml#L57-L68

[RegisHubelia]

Ah - thanks, I appreciate the quick response. Should of check the settings, so glad it's already an option.

[alexkorotysh]

@RegisHubelia
https://github.com/zammad/zammad-helm/blob/main/zammad/values.yaml#L66
set reindex = false to skip reindex each restart pod
If you need you can easily reindex from zammad pod

[RegisHubelia]

Thank you. This worked fine.