We have a P2000 G3 iSCSI array populated with 72 disks which we are using as a vSphere 6 replication target. The iSCSI traffic is switched through 4 dedicated 2920 switches.
The storage is carved up into a few different RAID10 and RAID5 LUNs. The physical drives are all SAS 15k 300Gb. Round Robin for the P2000 LUNs with 1 IOPS.
At the moment replication is turned on for around 20 smallish VM guests.
The source array is a NIMBLE which is not under any excessive load.
We are seeing intermittent errors for any given volumes on the P2000:
Lost access to volume 58cc4b53-78f6b5cb-812a-8cdcd412f1b8
(repltarvdi101) due to connectivity issues. Recovery attempt is
in progress and outcome will be reported shortly.
info
20/03/2017 19:52:20
repltarvdi101
Successfully restored access to volume
58cc4b53-78f6b5cb-812a-8cdcd412f1b8 (repltarvdi101)
following connectivity issues.
info
20/03/2017 19:53:42
esxvdi1.arb.co.uk
The switch ports all have flow control enabled on the 2920s.
I'm fairly confident (but not sure) that there is no bottleneck at the physical disks or network and my intuition is that we have a configuration issue.
I'm aware of the possibility that the P2000G3 may be just entirely incompatible, but I'd rather keep looking for answers for a while yet before I have to face up to that possibility.
Any ideas much appreciated.
thanks
