vsphere HA - host monitoing

[hypervisor01]

Question – in vSphere HA, when we disable host monitoring, does it stops check on datatsore heartbeat as well ?

The answer most probably is No, because if, host monitoring is disabled and while performing network maintenance, when the host is isolated, somehow the host crashes and go down, HA would then trigger failover.

As per my understanding, we use this option while performing network maintenance that may isolate the host. But even in normal scenario data store check is anyway done only after network heartbeat to confirm if it is an isolated host or a failed host. So it won’t reach upto that level.

[tomtom901]

Host monitoring (to my understanding) when host monitoring is disabled, this disables network isolation response and network partitions. When this is disabled, I think the host will skip the network heartbeat and go straight to datastore heartbeat. I'll see if I can confirm this (need to have a quick look at Duncan Epping's and Frank Dennemans book, I'm sure they have it) later today.

[hypervisor01]

That is looking to be right for me also. Please if you could confirm that, I am waiting. I have done enough reading but still I m not sure.

[abhilashhb]

Just checked it. under the test case of single host isolation this is the explanation.

When a host is isolated, the isolation is detected by the HA master node, as network heartbeats from it are not received any longer. When the master has detected network heartbeats are missing, it will start monitoring for datastore heartbeats. Since the host is isolated, it will generate datastore heartbeats for the secondary HA detection mechanism. Detection of valid host heartbeats will allow the HA master node to determine that the host is running but isolated from the network. Depending on the isolation response configured, the impacted host might choose to power off or shut down virtual machines or alternately leave the virtual machines powered on. The isolation response is triggered approximately 30 seconds after the host has detected it is isolated.

So yes, it will look for network heartbeat and when its not receiving any it will check the datastore heartbeat.

[tomtom901]

That appears to be without Host Monitoring. We are curious to see what is the procedure is when Host Monitoring is disabled.

[abhilashhb]

So this is what happens when the host monitoring is disabled. only network heartbeat will be disabled. the master drops the ownership of the datastores.

Quoting Duncan's blog article which says

Before the other hosts trigger the isolation response for a given VM they will validate if the datastore on which this VM is stored is “owned” by a master. In the case of a cluster wide isolation due to a network outage / maintenance the ownership would be dropped and this would result in HA not triggering the isolation response.

So i think datastore heartbeat will still be received.

[abhilashhb]

in another post duncan says

Duncan wrote a post in reply to this one where he recommended unchecking “host monitoring” instead of disabling HA. This will be quicker and make it to where your cluster wont have to start over. Moreover, the host monitoring option is really meant for outages like this. So don’t disable HA, just uncheck “host monitoring” during the expected outage window.

here are both the links

When to disable HA? « Cloud-Buddy

Re: when to disable HA? /cc @hashmibilal

[tomtom901]

Seems we all 3 were on the right page then!

[abhilashhb]

Yup. Seems so