Can the init logs time be configured?

[Yusuke1998]

It happens that I have implemented runboat in Kubernetes using Digital Ocean, and so far everything is fine; however, I notice that the init logs do not last more than 15/20 minutes, for some reason the associated job and pod are usually deleted on their own, I assume from the cleanup job, but I can't find a way to configure it, I have seen that in https://runboat.odoo-community.org/ they remain for much longer

[sbidoul]

Maybe this? https://kubernetes.io/docs/concepts/workloads/controllers/ttlafterfinished/

[Yusuke1998]

Thanks for responding. I congratulate you on your work; I think it's excellent.
And on that note, yes, I thought I'd go there, but I was also curious that https://runboat.odoo-community.org/ doesn't seem to have the same type of problem, as I saw some init logs there that were several days old. Could it be cluster-dependent?
I mean, some configuration in my case, Digital Ocean?
I tried adding ttlSecondsAfterFinished to the initialization job, but it didn't make any difference.

[Yusuke1998]

@sbidoul Greetings,
Apparently, the problem was that DigitalOcean, when managing Node Pools, in this case when deleting or creating a new one, also deletes the init containers from the build initialization job, leaving the init logs empty and returning a 404.
The solution for now was simply to increase the resources for the nodes; however, this doesn't seem like the best solution.

[sbidoul]

I congratulate you on your work

Thanks, much appreciated!

I the nodes disappear, the logs will disappear too, there is nothing we can do about it, given the runboat architecture which entirely relies on kubernetes for accessing the logs. You could of course setup a log capture solution to preserve all the logs of the cluster.