Hi,
This may be useful to someone someday.
I observed high IO delay (Datacenter > node > Summary) and hesitating containers. I had a failing disk in a ZFS mirror that repeatedly ignored (by issueing zpool clear) and then repaired (new disk and zpool replace). Only after resilvering the high IO delay went away.
Even while there only containers running on the boot drive (nvme ssd) and not on the ZFS mirror there were hesitations and hiccups. Only now, after the resilvering finished, IO delay is low (0.00% - 0.07%). CPU or memory were never (too) high.
I have a few Wordpress-containers were the load time of the front page seems correlated to the node IO delay.
cheers,
Ferenc
This may be useful to someone someday.
I observed high IO delay (Datacenter > node > Summary) and hesitating containers. I had a failing disk in a ZFS mirror that repeatedly ignored (by issueing zpool clear) and then repaired (new disk and zpool replace). Only after resilvering the high IO delay went away.
Even while there only containers running on the boot drive (nvme ssd) and not on the ZFS mirror there were hesitations and hiccups. Only now, after the resilvering finished, IO delay is low (0.00% - 0.07%). CPU or memory were never (too) high.
I have a few Wordpress-containers were the load time of the front page seems correlated to the node IO delay.
cheers,
Ferenc