712. system stack overflow. process terminated.

[mecheng]

Hi everyone,

I have a TI process that loads data from Oracle. It's been working great for about a year and a half now. Recently we're receiving a system stack overflow error while trying to run the process. Here are some more details of the model:.

TM1 10.1
Model is not very big. The data folder about 650 mb in size.
the cube we're loading into has a .CUB file size of 1,350 kb and feeders file size of 81mb. I've seen other stack overflow issues on this forum but they relate to Evaluation stack overflow. This is different. The IBM site tells me this happens if there are nested IF's and WHILE loops. however, my TI process does not contain these. My other TI processes that DO contain these still work fine. But they're pointing to a different cube.

My TI process uses dynamic views and subset creation (with cleanup in the epilog) but other TI's have the same concept and they run perfectly fine.
The server machine has 16gb of ram, running at 35% CPU usage and 3.4 GB of memory usage.
I tried restarting the TM1 service and the server machine but the problem persists.

Any ideas on this? what else can i try looking into?

Thanks,
MechEng.

[jim wood]

Have you tried running the data source query outside TM1? I ask as I'm trying eliminate the data base as the issue,

Jim.

[David Usherwood]

Your data load will be firing feeders in the destination cube. Try temporarily removing the rules (by renaming the RUX file and restarting) to see whether that cures the problem - if it does you can then study the rules and feeders.

[mecheng]

Thanks guy. i've done as you've suggested. removing the rule and loading data does indeed work.
so it narrows it down to a feeders issue. I'll study them and see what i can do.

[Wim Gielis]

While testing, you could also make use of the internal control cubes that TM1 uses for counting the number of feeders, for instance.
Make sure you bounce the server to make correct inferences.