Hosts losing and re-establishing connectivity to vSphere Datastores every 15 seconds

[hutchingsp]

We're in the process of having a new SAN (HDS HUS) installed and configured.

Long story short it's in a state where I'm able to have a bit of a play prior to them coming out to factory reset and commission it.

Our hosts are running ESXi 5.5 U2 and are directly connected via FC to the fiber ports on the HUS.

The FC HBAs are Qlogic 2562 dual port.

I've very quickly used the SNM2 wizard to create a volume and simply assign it to all the FC ports, then I rescanned in vSphere and sure enough there were LUNs which I then formatted as VMFS5 and set about creating a test VM.

The VM works perfectly but on the host under "Events" I'm seeing this about every 15 seconds for each LUN.

• Lost access to volume 5433dda2-86a51196-a348- 90e2ba4603a0 (Test 1TB) due to connectivity issues. Recovery attempt is in progress and outcome will be reported shortly. info 07/10/2014 21:02:56 vsphere01
• Successfully restored access to volume 5433dda2 -86a51196-a348-90e2ba4603a0 (Test 1TB) following connectivity issues. info 07/10/2014 21:02:56 vsphere01

Everything looks right in ESXi and some digging suggests this may be to do with datastore heartbeats based on the interval, but there is no mention of that in the Events log.

I've noticed it only starts once there are VM's on the LUN - it doesn't happen simply by rescanning then formatting.

Now I'm sure it's going to be something trivial and that when HDS come out and set the SAN up properly it won't do it, but does anyone have any ideas?

[Praveenmna]

This looks like connectivity issues.

Identify the storage array on this data-store is hosts and verify all data-stores on the array are having issues.

Also verify connectivity to storage switch and array are fine.

[warring]

Are the Qlogic HBA's drivers / firmware up to date? are the drivers installed on that ESXi build, eg IBM build or a HP build of ESXi. Checked the logs? vmkernal and so on.

[williambishop]

Don't know if this got resolved, but I ran into the same problem. The fix was to make sure vmware platform is chosen, and sometimes one of the ports will not take all of the check boxes, which need to be as follows:

You get there through host groups, edit the host group, then change options, about halfway down is a set of checkboxes, ensure these are checked on all ports the volumes are presented on.

Discovery Chap Mode

Unique extended copy mode

unique write same mode

unique comare write mode

standard vaai command mode

dp depletion detail reply mode