Mike Dame

What will be great about this is, once it's done, this whole repo will itself be an implementation of the framework (ie, the default Descheduler). So we should use and...

The examples shouldn't be dependent on what environment, as long as it's kubernetes :) but for me personally I also like Kind

I think a month is good enough soak time for people to see this :) As long as we have confirmation that there are no copyright issues (ie, you own...

As a reference for anyone who wants to pick it up, https://github.com/kubernetes-sigs/descheduler/pull/900 is an example of what we mean by `Also, to improve plugin modularity it's preferable to move corresponding...

/close done in #954

Based on the comment in https://github.com/kubernetes-sigs/descheduler/pull/781/files#diff-66db71b98f61a3c06ca8d17cbed99626fc23f8bab081cd71a8baf0867613d4d5R111, it sounds like the big question around this is whether to cache the list of nodes before the strategy runs, or pull them live...

As you pointed out, we do use the namespace filter early on in most strategies. Ideally, the evictor shouldn't even know about the namespace filters because all the pods from...

@migueleliasweb sure, thanks for offering to help. There's definitely something to be fixed here. Either we need to update the evictor checks, or ideally these pods should never even make...

ah yeah. it makes sense then to remove it to avoid that race condition. But we should add a follow up to add some small test to verify descheduler is...

For the [pods_evicted](https://github.com/kubernetes-sigs/descheduler/blob/a60d6a5/metrics/metrics.go#L36) metric, I'm not sure what the use case is for this. Won't you just end up with a bunch of new time series (one for each pod...