Issues with Valence jobs still being active after ending the tcp server

[Shawn Smalley]

We run a nightly backup of our system and before we start it, we end the tcp server for all of our valence instances, and then bring them back online once the backup is over. We have been running into an issue overnight where an operator will try to run a transaction after the system comes back up, but the transaction fails because a data area used in the transaction still has a lock held from one of the Valence jobs by user QTMHHTTP.

I noticed today that when i issue the ENDTCPSVR command, it immediately kills all of the jobs for valence aside from 4 jobs: all of them are for QTMHHTTP/VAL6LIVE and the functions they are running are 3 instances of VVFUSIONP and one of VVFUSIOND So i can only assume it would be one of these which could be holding the lock. Below is an example i recreated using our development environment.

image.png

Is there a way to guarantee that these get shutdown during our nightly process? right now I'm just doing a ENDTCPSVR SERVER(*HTTP) HTTPSVR(VAL6DEV) command for each of our instances and putting in a delay between each to make sure everything is finished before the next command.

[robert.swanson]

Hi Shawn,
Best to use the VVSTOP command (with the appropriate Valence instance library at the top of your library list), as that signals the Fusion jobs servicing that instance to terminate, in addition to the Apache jobs. You can then likewise use VVSTART to start everything back up.

[Shawn Smalley]

Perfect, thank you!