insufficient storage

[apparelbusiness.com]

Hello,

We had a situation where a program generated a hard error due to insufficient storage. The job has been running since 04/06 and had processed hundreds of requests and then this morning it generated this message. is there a way to prevent it from generating this error.


d10.png


These are the current settings.


w11.png

[robert.swanson]

There's no way to know from that screenshot what's apparently accumulating storage in the CGI job to the point that it throws that error message. It's possible the job just happened to reach its maximum allocation on what may have been a relatively small memory request for creating a JSON array response on a particular call, thus resulting in that exception.

In any case, this is likely stemming from one or more custom programs somewhere that need to be looked at -- my guess would be something isn't recovering allocated memory for the construction of a JSON response under certain circumstances. The activation group reclaim setting wouldn't have any bearing on memory cleanup in this case if the custom program(s) are running in the VALENCE or *CALLER activation group, so that might be one thing worth looking at -- seeing how the programs were compiled.

You mentioned the CGI job that threw the error had been running since early April, so another thing you might want to consider as a preventative measure is having that instance restart on some regular basis, like every Sunday night. This can be done with a simple CL program that you set up to be called from the job scheduler. The program should simply ensure the appropriate Valence instance library is at the top of the library list, then run command VVSTART. For most systems the instance restart usually completes in under a minute.

If you see this phenomenon happen again, and can leave the job in MSGW for a bit, give us a call or send us an email so we can get online with you to see if there's any obvious source for the problem.