TM1 Forum

I have been experienceing strange behavior that I have not encountered before. One of our models takes up about 2.5 gigs. In a few recent occasions when I have been dealing with rules/feeders I will save the rule and the server will then lock up. When I go the the task manager I see that the service is rapidly growing in size and will eventually eat up all the RAM (all 16 gigs of it). Once I restart the server it will reload to the 2.5 gigs and the rule that I was working will process correctly and will not grow out of control no matter how I work with the data.

What is causing this? I can work with the same set of rules all day long and it will happen 2 or 3 times turing the day when I save. I cannot pinpoint it down since sometimes it happens on a simple change where I am actually better defining the rule/feeder and not at times where I write a loosely defined rule/feeder that I would expect it to overfeed.

Any thoughts?

Hi George,

You may find that the server keeps the last working version of the rules. The actual text file itself keeps any updates. This has happened to me on several occasions. I have then gone back to a backup copy of the rules that I made and this kind of thing went away,

Jim.

Hi George,

TM1 evaluates rules (specifically feeders) differently during server load versus the actual saving of a rule file.

Chances are that you have a bad feeder that is causing a stack overflow. If this is encountered during processing feeders when the server loads it is kind of skipped over (sorry about the lack of tech speak.) However during saving of a rule file it will just keep churning and in your case consume all available memory. If the rule syntax is all correct then the new text of any modifications will be saved to the RUX file, it is just the processing of feeders in the commit stage that falls over. I suggest you check your rules and feeders carefully as this kind of problem is often caused by recursive rules. (do a search for "recursive" on the old forum) If this is the case you need to consider putting in some calculation break points.

Also check the server error log for tm1s stack overflows.

Damn Feeders, every time!