Our Setup is
5 Host cluster with shared storage. We have HA and DRS enabled
We had a Host failure with 4 running VM's all the VM's rebooted when they
migrated off to the other Hosts in the Cluster.
My understaning of HA was the images remained available when HA was envoked. Due to this we did some testing and changed the isolation settings to remain powerd on when a host was rebooted. This made no difference the VM's powered of again for 1min? I have read on the forum that das.isolationaddress can have some impact but not sure what that impact is. I would like some more info regarding HA configuration and if it is indeed possible to keep the VM's running when a host is rebooted.
HA in its current implementation restarts the guest when a failover occurs.
There are rumors that the next release will provide us with nonstop HA.
Message was edited by:
oreeh
Think about it: how could the running VM get transferred to another host in the cluster when the host where the VM was running on isn't there anymore.
Yeah that's correct. HA simply restarts the VM(s) on the standby server(s). It's like physically pulling the plug out of the host, how could it vmotion guests when its got not power (just one example of a failure)
Hello,
I'm not sure I totally understand your question but I will give it a stab.
If an esx host faults / becomes isolated, the vmguests are basically booted up on an available host in the cluster. For uplanned outages, this simply is the case, it is not vmotioned or anything like that, although it has been aluded that a future release could do something like this but only if the hosts are running in some kind of duplexed mode(not available now).
If the reboot of the host is planned, simply place the host in maintenance mode, and if the DRS cluster is in fully-automated mode, the the guests will move to another server. if it is in manual or partially automated, you will have to do the vmotions directly.
VMWare HA is really only a way to start up guests on a new host when one of the hosts has a problem.
Hope this was helpful.
Regards,
Jon
You're not going to be able to vmotion with a server failure as it is the equivalent of clicking reset on the VM and powering it off.
If you're wondering about rebooting the host in a controlled manner, if you have DRS setup, then the host will vmotion the VMs off the server when you put it into maintenance mode and the server can then be rebooted.
Why is there an option to "remain powered on" on the isolation settings what does this have an effective on
If the host finds that it is isolated (i.e. it has a failure but is still running), the default option is to power down the VM. You can change that so that the VM stays powered on.
Here's a thread that discusses the options
http://www.vmware.com/community/thread.jspa?messageID=605107
With some VMs it's safer to not simply power them off.
Database servers are an example of this.