DataCenter resource allocation suggestion

[Tong Wang]

Hi everyone, 

Due to the end of the server announcement, we are planning to move to DataCenter and use K8s to manage the whole cluster. So my question is how can we allocate resource for each pod? 

Current: both Jira and COnfluence are in m5.2xlarge instance, and set jvm from 4G to 9G for both Jira and Confluence

So for the new infrastructure, how many pods we will be needed, and for each pod what should be the spec? 

Current server usage: 

Any suggestions would be highly appreciated. 

[mogavenasan]

Hi @Tong Wang,

I'm not sure what you are referring to as "pod", I will assume that you are referring to nodes. You can run a single node Jira Data Center if you want. However, if you would like to enjoy the high availability that comes with Jira Data Center, then yes you will need a Load Balancer and more than 1 Jira Data Center node.

Your golden question is - how many nodes you need?

Well, it depends on the load, if you are using m5.2xlarge instance and we can see the CPU load is not even going up to 50%, then it means that you do not need more than one node.

If you want to use multiple nodes, then you can downside the instance type to something smaller and have 2 nodes cluster. Once you have 2 nodes, then you can monitor their resources, if needed, then you can add one more. 

For memory utilization, you will need to look at the Java Garbage Collector log. The machine memory usage is not sufficient to indicate if the heap space is doing good for Jira.

I hope that this helps.

Thanks,
Moga

[Pramodh M]

Is it in AWS?

[mogavenasan]

AFAIK, m5.2xlarge instance type only exists in AWS. 

[Pramodh M]

Yes, Great thought on the suggestion though!

Thanks, @mogavenasan