I did something similar myself but with a few differences. I tested out this solution and found that it works for 1 instance but not for multiple... helm install so -n instance1

helm install so -n instance2 --set global.nodePortPrefix=303 Error: release instance2 failed: persistentvolumes "so-mariadb" already exists

helm list NAME REVISION UPDATED STATUS CHART NAMESPACE instance1 1 Wed Mar 7 19:05:53 2018 DEPLOYED so-2.0.0 default instance2 1 Wed Mar 7 19:07:00 2018 FAILED so-2.0.0 default

get pv --show-all NAME CAPACITY ACCESS MODES RECLAIM POLICY STATUS CLAIM STORAGECLASS REASON AGE so-mariadb 2Gi RWX Retain Bound instance1/so-mariadb 10s

The strange part is that you can delete the failed instance, create it again and it will come up but it will steal the volume mount from instance1!

kubectl get pvc --show-all --all-namespaces NAMESPACE NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE instance1 so-mariadb Lost so-mariadb 0 4m instance2 so-mariadb Bound so-mariadb 2Gi RWX 2m