By default, vRealize Operations Manager alerts us when OS partition/disk is out of space. But sometimes want to exclude some disks or partitions from such alerts because they are full by design. For example, you have a separate partition for pagefile, rotated logs, or just userspace that can be full, and it’s ok to us. This post will show you how to exclude a specific partition, let’s say /home/user, from alerts. It’s easy, but from my experience, I can see that people often don’t know how to do that.
Create a new symptom with disk exclusion:
Go to the Alerts tab, then click on Symptoms Definition, and create a new one.
In the second step, on the right site, select:
– Base Object Type: Virtual Machine.
– Symptom Type: Metrics
Then expand the Guest file system section and double click the metric that you want (most common: Partition Utilization (%) or/and Guest File System Free (GB). If you do not have that second metric that I use here, read the post to the end, and later I will explain how to enable that metric.). Remember that it does not matter which partition you choose here. Why? I will explain that a little bit lower.
NOTE: What to do if you can’t see the partition you need, and you are sure that one VM has it? Or you can see all Linux partitions, and you need Windows? Just click on the Select Specific Object button and find VM that has a partition that you need.
Now, we can configure thresholds, make symptoms to monitor all partitions, and exclude the partitions that we want.
In my example, I would like to trigger a symptom if partition space is less than 1 GB, and it must be critical severity.
And here we come to the most critical step. Select Evaluate on instanced metrics checkbox!. It means that the vROPS will be automatically monitoring all instances of that metric. So, every partition that the virtual machine has will be monitored by this symptom. Even if you create a new partition on OS, it will be added to that symptom.
Enable Evaluate on instanced metrics allows us to add partitions to exclude the list. So, we want to add /home/sftpuser here. It means that all partitions are included in that symptom except these on exclude list.
NOTE: That part usually confuses people because, in symptom definition, we can see a specific partition name but evaluate on instanced checkbox makes that it does not matter…all partitions are assessed to monitor.
Create a new alert:
We have a new symptom, and now we need to use it in the alert. So let’s create a new alert and enable it in the policy (If you want to use only your custom alerts, you should disable built-in alerts that also monitor disk space on VMs in Policy).
Note: If you want to exclude partitions from alert only for the group of the virtual machines, you need to create a separate policy for that VMs (for example, grouped in Custom Group) and enable that alert only in this policy.
Drag and drop our symptom here.
In my example, I do not need any recommendation, so I leave that part empty.
Remember to enable alert in the policy that contains virtual machines which you want to monitor using that alert.
Optional: Enable metric: ‘Guest File System Free (GB)’
To enable Guest File System Free (GB) metric. You must go to Administration tab -> Policies, and edit your policy like in example below:
All done. Simple configuration and you can avoid some false alerts. Have fun.