cgsLauncher.exe process at 100% CPU usage

[Dev1975]

Hi all,

My Customer is currently experiencing high CPU usage by the cgsLauncher.exe process on the ContentManager server. This started around 9am on a Monday and has not lowered since. I am currently awaiting further information for review but have done some research into this process. Unfortunately, information on this process is very thin on the ground and the few IBM articles I've found on this issue suggest upgrading to the latest version, which doesn't apply as the Customer environment is already installed and patched with the latest BI release.

Has anyone come across this on 10.2.1 FP4? If so, what was the cause and how did you go about resolving it? I'm hoping that this can be fixed without having to do an environment restart!

TIA,
Dev

[Karti]

Have they done a reboot of their server(s)/Cognos service(s)?

[bdbits]

Have you called support? Seriously, I love this site but would not rely on waiting for replies here for something affecting a production environment.

[Dev1975]

Update: The CPU is no longer running at 100% usage and is now at lower levels. I have been monitoring it and it looks like it only spikes during the morning when a lot of people are logging on and using the environment concurrently. Once they have stopped using it, the CPU usage goes down.

Does anyone know of any trace files that can be enabled to see what is causing these spikes? The cogserver files are overwritten quite quickly (logging levels are quite high) so any useful info is over-written before I have a chance to view it. I am keeping a close eye and will try and grab the logs when it happens again.

There is precious little information about the cgsLauncher process out there, which is frustrating. Any tips/info gratefully received!

[Jaro2]

We experienced the same behaviour. This was because of a graph with a lot of barswith a value of 0. With showing values on the bars, the process went up to 100% for days and cognos not showing the report.

Solution, do not choose the option 'fine' for showing values in a graph!

[sdf]

How many users are using the system during the peak hours (in the morning)?
This may relatively be caused by low memory set for Java Virtual Machine (Maximum memory for Tomcat).

[Dev1975]

Hi Feds,

Not sure how many people are accessing the system but user load is definitely the likely culprit here. The environment is deployed to Websphere and I'm currently trying to get the JVM Heap Size info. I'll paste it up when I get it.

Cheers,
Dev

[tshw]

We experienced the same behaviour. This was because of a graph with a lot of barswith a value of 0. With showing values on the bars, the process went up to 100% for days and cognos not showing the report.

Solution, do not choose the option 'fine' for showing values in a graph!

Hi Jaro2,

We ran into same issue with the graph defaulting to fine staggered.. then when pulling chart into Workspace, it was pegging CPU to 100% as well.

I think this is a bug, since this is dangerous for business users when using charts and dashboards. Don't you agree? We're trying to get IBM to admit this is a bug and not an enhancement request.

Thanks..

[Dev1975]

Hi guys,

Not sure if this was made known but the cgsLauncher hogging CPU was a bug introduced by the Fix Pack installer. This was cured by installing FP5 into the environment.

Hope this helps.