[bitnami/etcd] Scale from 0 or 1 for replica lost data

[alisre]

Name and Version

bitnami/etcd 9.6.0

What architecture are you using?

amd64

What steps will reproduce the bug?

1. Start new etcd node helm install etcd-test bitnami/etcd --set auth.rbac.create=false --set replicaCount=1 --set persistence.enabled=true --set global.storageClass=disk
2. Wait the etcd installation is finished
3. write the data to etcd etcdctl put foo "hello world"; etcdctl get foo hello world
4. Scale down to 0
5. Scale up again to 1
6. etcd running and get foo but nothing etcdctl get foo

Are you using any custom parameters or values?

auth:
  rbac:
    create: true
    allowNoneAuthentication: true
    rootPassword: "abc_@%yRGxGV0Y_%@ABC"
  token:
    enabled: false
    type: simple
autoCompactionMode: "periodic"
autoCompactionRetention: "24"
initialClusterState: "new"
initialClusterToken: "etcd-cluster-k8s"
removeMemberOnContainerTermination: true
extraEnvVars:
  - name: ETCD_MAX_REQUEST_BYTES
    value: "10485760"
  - name: ETCD_QUOTA_BACKEND_BYTES
    value: "8589934592"
  - name: ETCD_HEARTBEAT_INTERVAL
    value: "150"
  - name: ETCD_EXPERIMENTAL_BACKEND_BBOLT_FREELIST_TYPE
    value: map
replicaCount: 1
podManagementPolicy: Parallel

What is the expected behavior?

etcd will not lost data

What do you see instead?

etcd lost data

[aoterolorenzo]

Hi @auto-sre,

Indeed I could also replicate this behavior. As it doesn't seem a trivial issue I will raise an internal task for the team. We will reach you back on this issue at the time we are able to address it.

[alisre]

ok thanks

[zhuoyang]

any updates on this?

[carrodher]

Unfortunately, and due to other priorities in the team, we were not able to work on this yet. We'll update this ticket once we have more news.

Having said that, if you would like to contribute with a fix, PRs are welcome!