You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Another team uses the unsupported IPI method to deploy their clusters, and are using qcow2. I wonder if applying the pattern in this tuning playbook would be sufficient to avoid the iowait issue, and continue to use qcow2?
We've observed very high iowait time being consumed by OCP cluster nodes when they are performing high amount of disk IO.
Please update the automation to give the option to use raw block (SCSI LUN or DASD) devices and pass these through to the OCP guest nodes.
The text was updated successfully, but these errors were encountered: