Change Thanos PV Access Mode

Note

This procedure was applied in 2020 when some Thanos were already deployed but wit a wrong access mode.

1/ Change block accessModes from ReadWriteMany to ReadWriteOnce in the configuration file

Note

The changes will affect the configuration of persistent volume claim (pvc).

2/ Change access mode to pv

This procedure apply to Thanos and Prometheus persistent volumes.

$ kubectl -n monitoring get pv # to get pv name
$ kubectl -n monitoring edit pv <pv_name>  # change accessModes from `ReadWriteMany` to `ReadWriteOnce`

3/ Delete thanos-monitoring-compactor statefulset

Modifications to statefulsets are allowed, only if statefulset is deleted.

$ kubectl -n monitoring get statefulset # to get the name of the stateful set
$ kubectl -n monitoring  delete statefulset <statefulset_name>

4/ Redeploy monitoring stack

Note

To see more information about the zone see Logging namespaces nomenclature on Caascad

5/ Check if pv and pvc have the same access mode

$ kubectl -n monitoring get pv, pvc # to get persistent volumes and persistent volume claim names
$ kubectl -n monitoring edit pv <pv_name> 

Important

Before deployment, the pv and pvc are in mode ReadWriteMany (several nodes can attach a pv in read-write mode). After deployment, the pv and pvc are in mode ReadWriteOnce (pv can be attached in read-write mode to a single node).