All committers have signed the CLA.

@simonswine I reviewed the change which has different with our Rust Mode Setting. For helm installation, RustFS provides two mode

• Standalone mode: one pod with one pvc, namely SNSD(single node single disk mode)

• Distributed mode: MNMD(multiple node and multiple disk)

  • 4 pods with each pod has 4 pvc, totally 16 pvc
  • 16 pod with each pod has only 1 pvc, totally 16 pvc, so the below code

          {{- if gt (int .Values.volumesPerReplica) 1 }}
         {{- range $i := until (int .Values.volumesPerReplica) }}
         - name: data-rustfs-{{ $i }}
           mountPath: /data/rustfs{{ $i }}
         {{- end }}
         {{- else }}
         - name: data
           mountPath: /data
         {{- end }}
  

  means if the replicas is 16 and the pvc will also be 16, and totally pvc will be 16*16=256, that's not our exception.

I see what you are saying, I can fix this with a bit more defaulting, but what I am confused with is, what if I wanna run 8 replicas with two volumes (or any other number than is not 1-1, 4-4, 16-4 [replica-volumes]), is that be bad for performance or just right out broken?

@simonswine Good question, actually, 8 replicas are also fine, even other replicas( for example, 32 64) are also fine. As you know, MNMD mode has several node and disk combination, but helm chart can not cover all the combinations, so we just support two main combinations( 4 replicas and 16 replicas) which is used frequently. If the current helm chart can not satisfy the user's requirements, user should modify the helm chart locally, but we will not accept the change.

@simonswine Hi, could y please fix the conflict?

@simonswine Could y pls help to handle the conflict?

@majinghe

@simonswine Good question, actually, 8 replicas are also fine, even other replicas( for example, 32 64) are also fine. As you know, MNMD mode has several node and disk combination, but helm chart can not cover all the combinations, so we just support two main combinations( 4 replicas and 16 replicas) which is used frequently. If the current helm chart can not satisfy the user's requirements, user should modify the helm chart locally, but we will not accept the change.

@simonswine Could y pls change the code according to this logic?

@houseme is our administration, he fix the conflicts, but i think the change code don't satisfy our requirement. If there is no any update, we will close PR due to it hangs on a long time. Thx.

@simonswine Due to the PR, there hasn't been any update for more than 3 weeks. We will close this PR. If you want to continue contributing with this PR, you can reopen it or create a new one. Thx again for you contributio.