Thin Disk Monitoring

[scotty_p]

We're running ESX 4.0 and want to start building some VMs with thin disks. We use 3rd party monitoring (ipMonitor) to monitor for disk space utilization on our VMs. I'm wondering how/if it's possible to monitor for disk space on thin disks. If they can grow to any size they want how will monitoring software know where to alarm?

Thanks,

Scott

[weinstein5]

First off the virtual disk cannot grow to any size it is capped by the disk size you wet when it was created - the important thing you need to monitor is the space on the VMFS datastore when using thin disk provisioning and this can be done using the vCenetr alarms - sending alerts, both SNMP traps and email alerts, when the VMFS datastore has a certain percentage of available disk space -

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful

[scotty_p]

Thanks for the response. We do monitor datastore usage, but we're also monitoring C:, 😧 drives on VMs.

Let's say you set up a thin disk with 20GB space and had your monitoring software alarm at 10%. Then your VM passed that 20GB. Is there a way to monitor that other than manually checking or will monitoring software alarm at 10% and stay in alarm?

Thanks,

Scott

[weinstein5 · Accepted solution]

If you have thin provisioned disk for your C and D drives both sized at 20 GB - neither virtual disk will ever exceed 20 GB - or take more than 40 GB of disk space - what you have to worry about is if you overcommitt the VMFS datastore - say you have 6 of these machines sitting on a 200 GB VMFS datastore - each with a C and D thin provisioned disk sized for 20 GB - initally each disk only has 5 GB of disk spaced used - so initally each VM when powered off will consume 10 GB for a total of 6 x 10 GB or 60 GB of the VMFS Datastore - when the VMs get powered on a per VM vmkernel swap file is created equal to the amount of memory assigned to the VM minus the reservation - so the machines are running and as data is added to each VM the thin provision disks will grow to store that data what are the consequences of the VMFS datastore running out of space -

1. If a VM goes to right additional data and there is no space available on the datastore the VM's OS most likely will through a write error and possibly crash

2. If on of the VMs is powered off and you try to power it on and there is not enough space to create the per VM vmkernel swap file it will not power on

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful

[scotty_p]

OK. That makes sense. So basically you just have to keep an eye on them if you're overcommitting - but you can't ever go over the specified size of the disk anyway. I set up a test VM and monitored the disk space as it grew. I set it at 10GB and it went into alarm when it reached 9GB.

Thanks,

Scott