Hi
Matthias,
Without knowing what activity triggered the spike in memory it's impossible to determine root cause. Assuming a TI process wasn't running at the time of the crash the mostly likely activity to cause this sort of problem is a large MDX view being processes. If the problem could be reproduce with MDX debug logging enabled that might help us understand the problem. It would also give us something to test against both 2.0.9.10 and 2.0.9.11.
https://www.ibm.com/support/pages/how-enable-mdx-debug-logging-tm1-serverI have not seen a case where the TM1 Server crashes with an out of memory error when there was still available system memory. Did the error in the tm1server.log file indicate how much memory was used? Can we see the exact error?
------------------------------
Stuart King
IBM Planning Analytics Offering Manager
------------------------------
Original Message:
Sent: Fri February 04, 2022 03:59 AM
From: Matthias Mazaj
Subject: TM1 Server OutOfMemory
Hi Stuart,
thanks for the response. I doubt that the actual limit was reached within a few seconds, as these would mean that the TM1 Server would need to grab almost 55GB RAM. This solution is running for years and the only recent change was the update to a new TM1 Server Version. We never had this issue before but several times after the update.
We went back to a previous version TM1 Server ( Version 2.0.9.10 (11.8.00900.3) ) and it's running for a few days now without any issues. Again nothing was changes except the TM1 Server version. So I strongly believe there is a fault in 2.0.9.11 (11.8.010000.6).
BR
------------------------------
Matthias Mazaj
Original Message:
Sent: Mon January 31, 2022 02:23 PM
From: STUART KING
Subject: TM1 Server OutOfMemory
Hi Matthias,
I would guess that the system running TM1 actually ran out of memory. Assuming the memory use of a process spikes very quickly it might not appear on the memory monitoring tool. Most memory monitoring tools poll the memory utilization of the server on an interval (for example...every 30 seconds). Maybe the TM1 server process spiked (using all available memory) then crashed within the polling interval.
I would start by checking how often the the monitoring tool polls the memory utilization of the server.
------------------------------
Stuart King
IBM Planning Analytics Offering Manager
Original Message:
Sent: Fri January 28, 2022 03:25 AM
From: Matthias Mazaj
Subject: TM1 Server OutOfMemory
Hello all,
i got an issue with crashes due to "OutOfMemory". Naturally we thought the maximum RAM was reached, but that isn't the case. It barely reached 55% at the time of the crash:
The server crashed at the peak with the red arrow.
We are running the newest TM1 Server version 2.0.9.11 (11.8.010000.6)
I've checked several settings, like MaximumViewSize (is set to 2GB and no log entry indicates that this was reached).
Anyone have any idea? Maybe a bug in the TM1 Server version?
BR,
Matthias
------------------------------
Matthias Mazaj
------------------------------
#PlanningAnalyticswithWatson