Lost connectivity to storage - VM Behaviour

[warnox]

Hi,

I'm not able to find any information regarding what is the expected VM behaviour if all paths to storage (FC, iSCSI, NFS) are lost.

1. Will the VM still appear as online/powered-on in vSphere but be inaccessible (network and obviously console)?

2. When access to storage is restored, does the VM show up in a powered-on or powered-off state?

With reference to vSphere 5.5, in case this differs between versions.

Thanks for any help.

[haripadmam]

Hi warnox,

Based on my past experience, if all paths are down, VMs will go inaccessible and will powerup automatically once the connection is restored. I haven't encountered similar issue any recently.

[warnox]

Alright, I'm asking as I've only ever seen an outage of about 1-5 seconds and none of the VM's rebooted, so want to know how the state is maintained?

The only VM's I saw an issue with were in MSCS, where RDM's remained accessible but their OS VMDK's were on a LUN which dropped off for a few seconds. In a situation where a LUN is trespassed, I wouldn't expect a problem but clearly it didn't like something :smileyhappy:

[kastlr]

Hi,

VMware sometimes declares access to storage lost while this isn't always the case.

I.e., to test the path quality VMware sends a TUR (test unit ready) command down each existing path every 5 mins.

The IO itself has a TTL of 8 seconds (when I remember correctly), so if the response doesn't arrive within that time period VMware will declare that path to be dead.

Same could happen when a ATS Heartbeat check (introduced in ESXi 5.5 U2) will fail.

ESXi will than declare the VMFS as lost, but this might not be correct.

For some arrays the following solution did fix this behavior.

Enabling or Disabling VAAI ATS heartbeat (2113956)

Regards,

Ralf

[jagdish_rana]

Hi There,

If all path down then VM are  gey out (inaccessible) in vCenter console and I/O operation will stop.

Once all path are up get the VM back was to do one of the following:

    

       Reboot the ESXi host

• remove the VM from inventory, browse datastore, add to inventory
  • with this option, you would need to answer the ‘moved/copied’ question on next boot
• put the host in maintenance mode
  • DRS (you’re using DRS, right?) essentially does the 2nd option while not hitting the ‘move/copied’ issue

.

I hope it will helpful.

Thanks