I had this question twice this week and did a quick search on my blog and I wrote an article about it a while back, but I figured it wouldn’t hurt to repeat some of that and expand on it. I copied / pasted this from part from our book as I think it it spot on!
VM/App monitoring uses a heartbeat mechanism kind of similar to HA. If heartbeats, and, in this case, VMware Tools heartbeats, are not received for a specific (and configurable) amount of time, the virtual machine will be restarted. These heartbeats are monitored by the HA agent and are not sent over a network, but stay local to the host.
Although the heartbeat produced by VMware Tools is reliable, VMware added a further verification mechanism. To avoid false positives, VM Monitoring also monitors I/O activity of the virtual machine. When heartbeats are not received AND no disk or network activity has occurred over the last 120 seconds, per default, the virtual machine will be reset. Changing the advanced setting “das.iostatsInterval” can modify this 120-second interval.
I didn’t know that one. That is really cool. Thanks.
How many people have had success with this option?
Does this apply even if the version of the VM tools is out of date? There are plenty of customers where I have seen vmtools not being updated?
@Kyle: I don’t know what the adoption rate is to be honest.
@Harsha: Yes even if VMware Tools is out of date it will work, it is still recommended though to keep Tools up to date.
nice to know I was always worries when migrating ESXi version most of the time my customers take a lot of time before upating Vmware tools… and I was worried VMs were unprotected nice to know
I’ve thought about setting this up a few times, but I’ve always worried about the out of date tools – with the best will in the world, it’s hard to get all of them up to date all of the time.
It’s also a little annoying that a lot of appliance type VMs come with tools that you can’t really get in to update, or even not installed at all.
And it’s more than a little worrying that when you download and install a vMA, vCenter shows the Tools Status as ‘Not Installed’ – that’s just wrong 🙂
I have been researching the exact flow of the heartbeats, I know they go to the HA Agent on the host, but is vcenter needed? In other words if I enable VM Monitoring on my vCenter VM, will it restart if the system Blue Screens since the vCenter server is down?
@Vincent: No vCenter is not required. This also happens on a host level. So even vCenter will be restarted by VM Monitoring.
What if a VM appliance does not have VMware Tools installed. Will it be restarted if it does not have any IO?
Or does it trigger on a running to not-running state change?
Hi Duncan
If you have a VM in your cluster that doesn’t have VMware tools installed – is VM Monitoring disabled by default for that particular VM??
Give a try to VMware monitor from MindArray IPM (http://www.mindarraysystems.com/vmware-performance-monitoring-tools.php). Notifcations can be send out using e-mails or SMS and applying processing rules you can perform actions such as reboot, start ect triggered by alarms.
Hello I’m from Brazil and I follow your site
I would like to take away one doubt in a matter of monitoring the vm when you talk that to avoid false positives is also checked disk IO and network activity wanted to know if it already comes standard configuration monitoring or have to configure something specific to this .
I thank you.
I see this is an old post but many questions related to false positives are not answered. I am concerned that the vms which cannot get tools at all or cannot be updated will be constantly restarted if they have low CPU, etc. Not willing to test on my production setup I tried a standalone client but cannot find a way to turn heartbeat monitoring on when in free esxi without using vcenter.
You’re not talking about using this (discontinued) product?
http://www.vmware.com/products/vcenter-server-heartbeat