Nutanix Education Blog

Welcome to the Nutanix NEXT community. To get started please read our short welcome post. Thanks!

cancel
Showing results for 
Search instead for 
Did you mean: 
Community Manager

Tech TopX: User Defined Alerts

Consider you are a System Administrator of an organization who is responsible for clusters which are very sensitive to, for example, I/O latency. You want to be notified when any of your clusters' latency goes over a certain threshold. Right now we have a way to solve this by setting custom alerts based on your needs.

Welcome to Tech Topx. I am Tanwee Gandhi and I am an SRE at Nutanix. Today we are gonna talk about User Defined Alerts in Prism Central. Let's take a quick look.

Now we are logged into Prism Central. To create a new policy, click on the Gear icon on the top right. Click on Alert Policies. For a new policy, click on New Alert Policy on the top right. From the Entity drop-down, you can select a specific host or All Hosts, same with the VMs and Clusters. Today for this demo we are gonna select All Clusters, which are managed by this Prism Central. In the Metrics, you can again select Read IOPs, Write IOPs or any Read or Write Latency. Today we are gonna select the Average Cluster Latency.

We're gonna select the Impact Type as Performance for today's demo. We would like to have a critical alert whenever the latency goes over 1 millisecond, and a warning whenever the latency goes over 0.5 milliseconds. In the alert condition, you can select for how long the latency would continue for it to generate an alert. Today we're gonna select 0, hit Save. If you have an existing alert in your cluster, it would let you view and then you can decide to override the existing alert or go back to the original alert.

And now if you go to User Created tab, you would see a new alert generated here. Once the alert is generated, let's go to Alert page. Here you would see the user defined alerts, the custom alerts generated for two Clusters, the Chandra and ICARUS Clusters, which saves Impact Type as Performance.

I hope this was helpful to you and you would be able to apply this in your environment. If you would like more videos on any of the topics please let us know, and thank you for viewing.

1 Comment
Explorer

Wouldn't it be nice if the top 5 VMs were marked in Prism at points where high IO, b/w, or latency were above thresholds and/or 'normal' levels, so that we could retrospectively know which activity caused any historical peaks?