Tommaso Pozzetti

Hi @pebrc, thank you for taking a look at this. This is interesting, because I am fairly sure none of my nodes took longer than 5 minutes to rejoin the...

If I understand correctly, that issue should affect searchable snapshots in the cold tier. I do have some indices in the cold phase of ILM, but we do not have...

I just ran the same test on a brand new cluster with no ILM configured and just a handful of indices. I edited a node config in the Elasticsearch CRD...

@pebrc Thank you for reproducing and confirming! Looks like a combination of fixing that bug on the Elasticsearch side and possibly adding a configurable in the operator to override the...

Unfortunately at the moment I cannot easily test the new version, but if the timeout is now 5 minutes I am fairly sure that will solve our issues. I still...

Hi @squeed, thanks for taking a look. Here is a cpu pprof for 10s during which the service is inaccessible as the cilium agent is stuck in the loop of...

@squeed I am running it with the default value `16381`

@dtzar thanks for the info and details! I might have gotten confused then. Based on the following [line](https://github.com/kubernetes-sigs/cluster-api-addon-provider-helm/blob/v0.1.0-alpha.4/docs/quick-start.md?plain=1#L156) in the docs: ``` Remove the label nginxIngressChart: enabled from the workload...

@dtzar I think that would prevent the desired behavior of retrying until the first successful install. For example if upon initial install something were to fail due to some component...

@Jont828 Here is my use case: I create a workload cluster using cluster-api. This cluster will manage all its content through gitops. It will have a gitops controller of some...