Unload Cube - Feeders Now Do Not Work

[youngjedi]

I have unloaded a cube and now my feeders dont work. I have restarted the server and still my aggregations will not calculate. I'm fairly certain my feeders are correct since I checked them all before i did the unload. I am in 10.1/Architect. Thanks.

[asutcliffe]

I have unloaded a cube and now my feeders dont work. I have restarted the server and still my aggregations will not calculate. I'm fairly certain my feeders are correct since I checked them all before i did the unload. I am in 10.1/Architect. Thanks.

I wouldn't be so sure, if your model isn't fed properly at startup, there's something wrong with your feeders.

One point to note is that once a cell has been fed, it remains fed until the cube is unloaded or you restart your server (unless you're using persistent feeders anyway).

If you're editing the feeder statements and saving and re-saving your rules, it's possible to get the cube into a state where it's fed sufficiently but that some of those feeders weren't actually set by the feeder statements in the most recent version of the rules you have saved. Is this a possibility?

Are you using conditional feeders? Another option is that something in your model has changed such that a condition that was met previously is no longer true.

I hope that helps a bit.

[Harvey]

One other thing to watch is stack overflow errors. If you have a long feeder chain (say ~400 hops or so), the TM1 server can run out of stack space. Stack space is a fixed amount allocated when the server starts up, and hence when it runs out, there's not much you can do.

If you have stack overflow errors, they should show up in your server message log file. However, I have found quirks in TM1 10.1 that LOOK like stack overflow errors but don't show up in the log as such, so I'm not sure these errors are being reported correctly in the latest version.

If you find part of a cube fed and others unfed, this can be a symptom of a stack overflow problem.

When developing and testing, it's best to turn persistent feeders off, or you may not detect feeder problems. For instance, if you write a rule that over-feeds, then correct it or make it conditional, the feeders the original rule generated will remain.

Under this scenario, you could develop happily, never knowing you have erroneous feeders until you deploy to production, so best to be cautious!