[BUG] Scale down from cluster- to single-node-mode fails

[florian-scheurer-mw]

Is there an existing issue for this?

• I have searched the existing issues

Affected Version

affected Version 12.1.1

last tested working version: 11.0.0

Current Behavior

Reaching the state that graphDB helm-chart version 12.1.1 is running in cluster-mode (either by setting it up from scratch or upgrading it from an older version or by scaling up from single-node-mode).

• Scale down to 1 replica via helm-chart

• The "cluster-scale-down-job" runs through without throwing an error message

• The proxy pods are scaled down to zero successfully

• The node-pods are scaled down to one "successfully"
  Issue:

• The last node-pod "graphdb-0" is restarted, but it still expects cluster-mode and waits for other nodes

• This leads to unresponsive behaviour of the database.

Expected Behavior

All the steps mentioned in the "current behevior" until the "Issue:"

• Either keep the last node-pod running or restart it as needed
• The last DB-pod is configured to be a single-node database
• The database is still fully functional

Steps To Reproduce

Reaching the state that graphDB helm-chart version 12.1.1 is running in cluster-mode, either by setting it up from scratch or upgrading it from an older version or by scaling up from single-node-mode. (e.g. setting the replica value to 3)

• Scale down to 1 replica via helm-chart
• Check the DB-behaviour

Anything else?

No response