If you submit a djob and run simultaneously the operator will crash with a nil pointer dereference error. Submitting simultaneously is our use-case that unfortunately we cannot avoid.
The desired behaviour would be similar to the handling of secretscopes when the underlying secret does not exist. Rather than crashing, report an error in the logs and continue the next reconcile cycle.
To reproduce given the following run referencing a djob:
If you submit a djob and run simultaneously the operator will crash with a nil pointer dereference error. Submitting simultaneously is our use-case that unfortunately we cannot avoid.
The desired behaviour would be similar to the handling of secretscopes when the underlying secret does not exist. Rather than crashing, report an error in the logs and continue the next reconcile cycle.
To reproduce given the following run referencing a djob:
run:
kubectl apply -f job_and_run.yaml
output: