TM1 Forum

Our server is now suffering from an "Out of memory" issue. We are using 32bit OS and we understand that we really reached the limit during processing. I know one way to resolve this is to restart the tm1 server / recycle the memory but I don't want to keep on restarting this once everyday. Is there a way to schedule a restart of TM1 Server so it will be automatically restarted once a day? Please help. thanks in advance

Are you running as a service? If so, there's some ideas here: http://applixforum.olapforums.com/viewP ... adID=12802

If you're running as an application then no, there's no way to automatically restart.

HTH,
Martin

I'm sure it's in the link Martin posted but the most common way is to use a batch file called by the ExecuteCommand function in TI script.

Batch file very simple just:
Net Stop "tm1 service name"
Net Start "tm1 service name"

If the batch file is in the root directory of the tm1 database directory then only the file name needs to be passed without any path.

We had a play with the "Restart.bat" approach mentioned on that thread, on a test system. In my experience it was more reliable if you did a "SaveDataAll" first. I guess it's a timing issue, the batch file gave up waiting for the cubes to save and hit the start bit too early. Anyone else have a similar experience?

TJMurphy wrote:We had a play with the "Restart.bat" approach mentioned on that thread, on a test system. In my experience it was more reliable if you did a "SaveDataAll" first. I guess it's a timing issue, the batch file gave up waiting for the cubes to save and hit the start bit too early. Anyone else have a similar experience?

Yup, since we moved from applications to services a couplea-few months back. The very first time I did a shutdown (manual, not by batch file) I expected it to finish saving then shut down in a nice orderly fashion.

It didn't happen that way.

If all of our cubes need saving then it can take 5 or 6 minutes to finish the process. Long before then I got an error 1053 (from memory) - something along the lines of "the service did not stop in a timely fashion". At that point it just left me with a half-saved .cu$ file from the cube that it was working on at the time. When I cleared the dialog the service started up again but of course it had to reload everything from the log files.

Ever since then I've just made sure that I've done a data save first. Being on 8.2.12 using SaveDataAll in a TI isn't an option for us since it will periodically freeze your server. Accordingly I haven't experimented with using Net Stop.

It does sound like your problem was similar to mine; it may not have been that the Start happened too early but rather that the Stop gave up on trying to save the cubes.

I do recall (dimly) reading something in Windows documentation about how you can increase the wait time for services, but it hasn't been a high enough priority for me to try to track it down again. If I find it, I'll post it.