Server holding file locks after job ended

[edward.thomas]

I'm finding that some of the Valence server jobs are holding onto a few file locks after the jobs they processed finished. The call stack only contains Q*** jobs yet there are vv*** Q*** and some of the program files still open.

All the RPG's end with Seton LR, Return and files are under program control. There is no log info to make me think they end abnormally in any way.

[robert.swanson]

This is a normal behavior of the CGI jobs to improve performance between calls, particularly for anything running under the VALENCE activation group, since that activation group never terminates.

If the open data paths are causing you an issue, try using a custom activation group name (something other than VALENCE) in your programs. Many sites make a copy of the VVHSPEC copy source, calling it something else like VXHSPEC, and changing the ACTGRP to something resembling their company name -- i.e, actgrp(ACMECORP).

If you're still seeing open data paths -- and they're causing you a problem for some reason -- you can force activation groups to be reclaimed after every call. See "Activation group to reclaim" and "Activation group reclaim interval" settings in Portal Admin > Settings. There may be a slight performance hit for doing this, so you should only go this route if the open data paths are causing problems elsewhere.

FYI, some good information on how this ODP process works can be found in this Code400 forum post from Birgitta Hauser and this MC Press article from Joe Pluta.