Hey Guys,
I have been hunting over numerous threads and blogs trying to find an answer to this question but cannot seem to find an answer, so hopefully somebody here will be able help me out with this.
Basically my scenario is I have 2 ESX hosts in seperate buildings, building A and building B. Each host is directly connected to a iSCSI switch and those iSCSI switces are directly connected with a fiber link in between buildings. The SAN is in building A and is only directed connected to the iSCSI switch in building A.
The issue I am having is if the iSCSI switch in Building B goes down, the VM's do not failover to the fully functioning host in Building A, they simply become unresponsive (obvisouly) and just hang until the iSCSI communication is restored. I have been testing this by pulling the cables from the iSCSI switch in building B.
Any suggestions are greatly appreciated and if you need any additional information please let me know.
Thanks,
I realise that is how it works but is there not a way to have it detect this and failover, maybe by creating an event/task for thwn both NICS go down..
If you are referring to the lock file on the VM, does this not timeout after a few minutes? Unless you're referring to the actual access to the SAN but this will be possible because HOST A would still have communication with the SAN however HOST B would not.
Would it be possible to have the vCenter heartbeat go over the iSCSI network, and if that heartbeat is lost the host assumes it lost communication?
Alright just a bit more information, I have performed some additional testing and it looks like when the VM is in FT and I pull the iSCSI connection it will instantly failover however I still cannot get this working with simply HA. I assume since the host can detect this failure with FT enabled it should be smart enough to detect this failure with HA enabled and restart the VM's on the other host.
Thanks,
I would think HA would start it up on HostA if the lock on the VM times out which is another thing HA Checks for besides the Network and CPU activity to see if it can see if the VM is still running. What settings do you have setup for HA? Do you have Host Monitoring turned on? What VM Monitoring Level do you have Enabled? VM Monitoring Only? and what is the Monitoring sensitivity that you have set? Low, Medium, High? Also, do you have the VM set to shutdown, powerdown or stay running?
My HA settings are as follows:
Enable Host Monitoring
Admission Control: Enabled
Admission Control Policy: Percentage 25%
VM Restart Priority: High
Host Isolation Response: Shut down
VM monitoring: Disabled
Have you tried switching the VM Monitoring to VM Monitoring Only and setting the Monitoring sensitivity to High?
Yes and I have seen no change.