VMware Cloud Community
scooter_
Contributor
Contributor

Simulate VM Failure to Test HA?

I have two ESXi 5 Hosts in a cluster.

Host A : Running (Master)

Host B: Connected (Slave)

Each of the WIn2008 R2 VMs Diaplay: Protected

So I think that I have the VMs in the right cluster and configured, though I'd like to Test the HA and force a failover to the other host.

What is the best way to do this test? What is the procedure for bringing the Original Host back up and having it run the VM it originally was running go back online?

What can be setup to Alert someone (via email or) that a Host has gone offline?

Thanks!

Scott<-

Tags (2)
0 Kudos
4 Replies
vmroyale
Immortal
Immortal

Note: Discussion successfully moved from VMware ESXi 5 to Availability: HA & FT

I think the best test is to simply pull the power from one of the ESXi hosts.

Alerts can be configured in vCenter alarms to send e-mails.

Brian Atkinson | vExpert | VMTN Moderator | Author of "VCP5-DCV VMware Certified Professional-Data Center Virtualization on vSphere 5.5 Study Guide: VCP-550" | @vmroyale | http://vmroyale.com
0 Kudos
sparrowangelste
Virtuoso
Virtuoso

what about sending a panic to your kernal?

use: vsish -e set  /reliability/crashMe/Panic 1

that would simulate a pscreen

Here is my test:  http://sparrowangelstechnology.blogspot.com/2012/08/psod-getting-purple-screen-of-death-for.html

--------------------- Sparrowangelstechnology : Vmware lover http://sparrowangelstechnology.blogspot.com
0 Kudos
scooter_
Contributor
Contributor

I thought about pulling the plug, though waht about just pulling out the Network Cables? That would simulate a Switch Failure. What happens when the 'Switch' is Fixed and now the 2nd Host has started the VMs and the original Host is back online and already has the VMs running? Will the 2nd Host know to shutdown the VMs that it started?

The forcing of a purple screen  or yanking the Power scares me. (-;

Thanks!

0 Kudos
vmroyale
Immortal
Immortal

Datastore Heartbeating in vSphere 5 complicates the old approach of pulling the mgmt network connection. Ideally, you would want to test each failure scenario (physical connection, storage, etc) of the cluster to see/understand how it will work so pulling the mgmt cable would also be a great test to perform.

Brian Atkinson | vExpert | VMTN Moderator | Author of "VCP5-DCV VMware Certified Professional-Data Center Virtualization on vSphere 5.5 Study Guide: VCP-550" | @vmroyale | http://vmroyale.com
0 Kudos