Onpremise kubernetes for atlassian products

Question

Hi,

We are trying to install on premise (data center version) a Kubernetes cluster to support all atlassian products.

Mind that we are not using AWS / AZURE or GCP product but vmware hosted VMs using flat-carr OS for our K8S clusters

We are hitting a nail with JSM + lucence's no NFS for its indexes. We use NFS for kubernetes for pretty much everything on K8S So you can see the problem.

What are you guys using for the "local side" needed by Lucence's need for indexes.

Any input / stories are appreciated :)

Thank you

Yevhen_5guu4k · Answer

Oh, on-prem K8S is like that. If you were in AWS I'd say EBS right away. In VMWare... no idea honestly. Whatever block storage is available. I'd probably look at things like https://github.com/openebs/openebs but honestly it's been ages since I last time had to administer an on-prem K8S cluster.

Jean-François Couture_lm63ua · Answer

We are using NFS for shared home but local home is non-standard to put data directly on the K8S Node.

As per support ticket, Atlassian support team confirmed that you cannot use NFS on local home as this does not work with lucene's Indexes on NFS.

trying to find another solution for local home that is not K8S node's local disk.

Yevhen_5guu4k · Answer

@Jean-François Couture what's the exact problem? Do you use NFS both for shared home AND local home and it impact performace?