We assume we have
- an active cluster
- a distributed job chain
- e.g. the first node is stopped
This node state is stored in the database table SCHEDULER_JOB_CHAIN_NODES where no special CLUSTER_MEMBER_ID was written in the row.
Now the above distributed job chain is changed to a "normal" job chain
then the above stopped job chain node is unstopped for a while and then stopped again.
If this job chain node is unstopped manually again via e.g. JOC then it is stopped after while automatically.
It seems that the above row in the table SCHEDULER_JOB_CHAIN_NODES without a specific CLUSTER_MEMBER_ID beats other rows with a specific CLUSTER_MEMBER_ID.
The state of a job chain node should not changed automatically.