Release a processing order --> job step is started, but never finishes due to problems on Kubernetes side
Processing order is suspended --> order is "SUSPENDING", job is "ON_HOLD", job step is still "RUNNING"
After correction of the problem in Kubernetes, the hanging job step is forcibly suspended --> job step now "INITIAL":
It was expected that the job now reverts to status "INITIAL", since no more job steps are running, and the order hence changes to "PLANNED", since no more jobs are "ON_HOLD", but this is not the case:
Workaround: No workaround in CLI or GUI is possible, since orders in status "SUSPENDING" and jobs in status "ON_HOLD" cannot be updated. It was necessary to change the status values manually in the database tables.
Implemented with a bottom-up propagation from job step state to job state to order state (commit f436094). Tested during S5P DDS6 reprocessing campaign.
The following sequence of actions was executed:
It was expected that the job now reverts to status "INITIAL", since no more job steps are running, and the order hence changes to "PLANNED", since no more jobs are "ON_HOLD", but this is not the case:
Workaround: No workaround in CLI or GUI is possible, since orders in status "SUSPENDING" and jobs in status "ON_HOLD" cannot be updated. It was necessary to change the status values manually in the database tables.