Conversation
| # | ||
| # Ingress / Router | ||
| # | ||
| # HAProxy router timeout |
There was a problem hiding this comment.
I'd also note to remind customers to check for other upstream proxies that may require overrides if just changing the HAProxy timeout doesn't resolve SSE issues.
| # OpenShift instances often have their own StorageClasses already created, | ||
| # instead of letting this Helm chart create one | ||
| # | ||
| # File count limits per PV |
There was a problem hiding this comment.
https://access.redhat.com/solutions/6221251
There are a few workarounds, here. It seems like this is inherited from upstream k8s. Curious that we haven't seen it on other non-openshift deploys
There was a problem hiding this comment.
I remember this one from The Bank, before the docs got paywalled, it was due to RHOSCP's thing with UID/GID, every time it attaches a volume to a node, it runs chown UID:GID on all files in the PV, which timed out > 1 million files.
There was a problem hiding this comment.
There's a couple workarounds in that docs to somewhat mitigate it, as well as an option added in 4.16
if the volumes affected by the long SeLinux relabeling are accessed via the ReadWriteOnce (RWO) access mode, and the cluster is running OpenShift 4.16 or newer, consider first using ReadWriteOncePod (RWOP) access mode in combination with the seLinuxMount option instead of this solution if the CSI driver used supports it.
trly
left a comment
trly
left a comment
There was a problem hiding this comment.
Looks good. My comments can be addressed if desired in the future.
2 participants
Checklist
Doc change.
Test plan
Verify with customers.