Hello,

As mentioned in this Github issue, Steve Hipwell and I propose to create to new endpoint on Jira Software Server/Datacenter.

An endpoint

/live

which sends back a 200 HTTP code every time the application is up and running, even during re-indexation.

An endpoint

/ready

which sends back the appropriate HTTP code if the application is ready for end users or not.

The problem is: while Jira is Full re-indexing, it sends a 5xx HTTP code to front-end (on both endpoints"/" and "/status_"_).
However Kubernetes watches for the application state and a 5xx HTTP code means "application down" and Kubernetes restarts it.
It's problematic when re-indexing.

Kubernetes good practices indicates that an application should have two endpoints as mentioned before because Kubernetes has readinessProbe and LivenessProbe mechanisms that doesn't fit well with actual Jira endpoints.

Today I used a workaround to watch out HTTP port but it is inaccurate compared to application endpoints.

Do you think that these modifications could be done ?

I declared component ADGS Design Changes, feel free to change it if needed.

Thank you