I don't have anything helpful to add. Except maybe make sure your I/O
scheduler is deadline or noop.
https://access.redhat.com/solutions/5427
We've been fighting a very similar issue here and have never been able to
resolve it.
Our system:
3-host vSphere 5.5 cluster
Three SANs (all RAID6)
2x Equallogic PS6100E
- 1 for storage of guest images
- 1 for replication
Equallogic PS6500E (42 7.2k 3-TB drives)
- data storage
We have a Linux NFS server with virtual disks are 8 12-TB VMDKs on the
6500E, joined via LVM into a single volume.
Average latency on some of the datastores is 100 ms, and we will
occasionally see up to 30-second latency in I/O (as reported by VMware).
https://imgur.com/a/mQM0H
_______________________________________________
Linux-PowerEdge mailing list
[email protected]
https://lists.us.dell.com/mailman/listinfo/linux-poweredge
