Ciro S. Costa

> it looks like this was a deliberate engineering decision made for the Helm > chart design. Will have to chat with @cirocosta when he is back Yeah, that's because...

> We believe that restarting pods using kubectl won't fix the problem of ATC > having an incorrect perception of workers' volumes; so we did the following: oh, actually, a...

### tl;dr Having both [remount volumes on restart](https://github.com/concourse/baggageclaim/pull/30) and [gc containers and vols present in the worker but not on the db](https://github.com/concourse/concourse/issues/3600), it seems sensible to me to consider going...

heey @vito, > These kinds of problems seem to be caused by not following that rule. ooh, totally! > it seems like we may want to either adjust the lifecycle...

From what I understand, the best case would be to have PVCs (so that the worker pod can have all the IOPS and storage to itself, regardless of where it...

slightly related: *Research Linux OOM killer behavior for cgroups #20*

While this is all great, ultimately, we should: - run this worker as a non-privileged container (otherwise, you can bypass any of these things with enough effort employed) - have...

I have the very same use case as @errordeveloper 👍 How did you end up working around it @errordeveloper ?

That sounds like a great idea @nsisodiya! The idea of the adapter seems to suit well as we have to be sure that the browser supports workers before trying to...

@kencochrane thx for the quick reply! That makes sense, got it. Would it be worth (and make sense) having more than one cattle of workers? For instance, say that we...