Open
Description
In a freshdesk ticket we understood that the participants of a workshop experienced something to be slow, but it never became clear what.
In such situation, it would be good to rule out it was related to the following:
- The node's network capacity (upload/download)
If many users would be reading large files from the NFS server or downloading from object storage at the same time for example, they may bottleneck in network capacity. - The node's ephemeral storage read or write speed
When a user in a pod on a node write files to something outside their home folder, they write to ephemeral storage on the node. - The node's ephemeral storage capacity (I think this may be referred to as disk pressure as a k8s node status)
Nodes doesn't have infinite capacity. If we run out, I'm not sure how we would error. I think we may even get pod evictions but I'm really not sure.
Related
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment