Queue blocking on badly populated RAconnect statement

We experienced an issue with a job blocking the queue it was in entirely until the scheduler was restarted some time later, it did not show in queue monitor but no other scheduled jobs ran in that queue.

The problem appears to have been due to the following statement

RAConnect( agent, agentport, username, pwd )

all of the arguments are variables populated previously, notably the 'agent' one. That is pulled in using inifilegetkey from a config file but at the time the entry did not exist in the config file causing the issue.

The schedule.log reported the following
An error occurred while executing automation script: Line 10: Line 65: Remote Access Error: recv call failed.

I'm not concerned about the failure itself more the queue blocking behaviour. I've set the job up to run asynch in case it happens again for another reason

I think the remote connection error somehow made the job to hang. It is possible that one of the DLLs used in the middleware didn't unload and the job got stuck. To avoid this in the future, please set the job to run detached.

In case the job is already running detached and started from a service, it is possible the process crashed, but you couldn't see the system error that on the screen. The invisible message could also hang the queue, because from the queue point of view the process haven't completed yet and still running. You may want to check Windows event logs log for occurrences of "...Application…" error around the time that job failed.

In any case, please ensure the job is set to run detached.

FYI. The next maintenance release will support timeouts for detached script type jobs. This could also be helpful in the situations like your, because a job with timeout can be forcedly terminated in case it hangs.

Just case, version 3.4.33 is now available. This can help you to deal with stagnant script type jobs.