Matt Cary
Matt Cary
Ah, I wonder if it's https://github.com/kubernetes/kubernetes/issues/117070. Let me do some more spelunking. Thank you for the cluster versions you were running, that's very helpful. On Tue, Jul 16, 2024 at...
Yes, I think this explains the reason why both processes are killed: https://github.com/kubernetes/kubernetes/pull/117793 The change seems to have happened in 1.28 which explains your problems. On Tue, Jul 16, 2024...
/lifecycle frozen
We've found increasing the memory limit to ~500M works well, and hasn't seemed to have an adverse affect on other workloads on the node. @acondrat the memory resource can be...
See details for GKE mitigation and fix rollout in https://issuetracker.google.com/issues/338125229.
/hold I believe Uriel still needs to get kubernetes-sig membership before we can do this. We'll look into it.
> instead of just logging a warning after 30 seconds, we could implement periodic progress updates to give more visibility into how much of the file has been processed. Yes,...
How do you get the total number of files? That by itself is expensive and could take O(minutes)? O(hours)? on a large disk.
I think I like your idea of counting the files and updating progress from within the goroutine. I think that we do have to emit an event and not just...