Documentation for upgrading elasticsearch from archived repository to eck-stack

Elastic Stack Elasticsearch
1

Hi Team,

We are currently using elasticsearch on our kubernetes cluster installed using archived helm chart repository: helm-charts/elasticsearch at main · elastic/helm-charts · GitHub 8.1.3 version.
Now we need to update it to latest version but it is recommended to use eck-stack.
Is there any documentation for the upgrade elasticsearch installed from archived repo to the eck-stack.
Basically we wanted to use the same pvc and statefulset name while upgrading.
Right now i dont find anything in the docs by which we can pass or update the statefulset name which also gets appended to the pvc name, and because of that we are kind of stuck.
Please suggest something so that we can proceed.

2

Hi,
If any one could share some thoughts on this, Please share it.

Basically we wanted to keep the same statefulset name so that pvc name can also be the same.
Right now , i am not finding any provision where i can change the statefulset name.
Thanks

3

I do not use ECK or K8s, but I don't think such documentation exists.

I think you would need to snapshot your data on a NFS share or Cloud Provider, spin-up a new cluster in ECK, and then restore your data from the snapshot.

1 Like
4

Actually that process should not come when you are upgrading the elasticsearch , i think that should be used for any disaster recovery or if you want to restore some old data from the snapshot on different or same elasticsearch.

While upgrading we should use the same PVC, so there must be a documentation is that what i think.
May be someone from elasticsearch eck-stack maitainer can clarify this or may be let us know what is the process of upgrading elasticsearch from older charts to new eck-chart using the same statefulset name, pvc.

Thanks

5

Yeah, but you are not upgrading anything, you want to change the deployment from using Helm charts to using ECK, those are 2 different ways of deploying Elasticsearch.

Not sure if this is as simple as you want it to be, or even if it is possible, so the quicker and safer way would be to create an snapshot and restore it in a new cluster.

Maybe someone from Elastic can provide more feedback.

1 Like