Need help in elasticsearch snapshot

Elastic Stack Elasticsearch
1

Hi Team,

I am planning to migrate data from one cluster to another cluster by taking snapshot of indices. In current k8s cluster, I do not have shared storage which is needed to define repo location. We are using rook-ceph block storage for elasticsearch cluster.

My first question, is shared path necessary for taking snapshot?

Second question is, can i use one PVC with ReadWriteMany with 3 elasticsearch sts objects? If yes, how would i define addition pvc information in values.yaml. I already defined volumeClaimTemplate: template in values.yaml.

volumeClaimTemplate:
  accessModes: [ "ReadWriteOnce" ]
  storageClassName: "rook-ceph-block"
  resources:
    requests:
      storage: 200Gi

I appreciate if anybody could help me out in this regard.

2

Yes.

I can't answer your second one sorry.

3

Thanks for your reply. If use shared path and take the snapshot, is snapshot taken from each elasticsearch pod ( I have 3 replicas of elasticsearch pods) or it is taken only from only one pod?

Thanks,
Kasim

4

Snapshots are taken on each node that holds data that is part of the indices being snapshotted.

5

okay, if I define one path in esconfig.yaml file like path.repo: ["/es-backup"], then each pod sanpshot would be stored under /es-backup path. When other pods of ealsticsearch initiates the snapshot, would that not overwrite existing snapshot? I am clear with the snapshot concept. Could you explain me clearly

6

Each node only snapshots the data that it has, so there won't be anything overwritten.

7

for example, as I defined path.repo, each node creates separate folder under /es-backup and stores the snapshot. Am i right ?

8

Not exactly. There are directories created that map to the index structure. Then each node puts the data from the shards it holds into that structure.

9

so, all nodes stores the snapshot under /es-backup.

10

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.