Good afternoon all,
We’ve been having on-going memory related issues on our middleware server for a few months now. After around 2 weeks of usage our NodeManager seems to fail as the server runs out of memory. Normally a restart of the server is the fix however, with an increase in uses we’re worried this will need completing on a more regular basis.
Can anyone assist in understanding how to fix this problem? We’ve already increased the memory allocation on our managed server of the MWS to the following:
The lower limit is 04 GB and the upper limit is 06 GB for both Managed Servers. I suspect this was merely prolonged the failure and hasn’t resolved the issue.
Log file dump:
<May 25, 2023 2:27:01 AM BST> <SEVERE> <Fatal error in NodeManager server>
java.lang.OutOfMemoryError: unable to create new native thread
at java.lang.Thread.start0(Native Method)
Did you increase that memory on the Admin, Main, or Integration server instance? I ask because node manager is in the Admin server area. How many active users do you have. I usually have between 100 and 350 and 2GB is all I need.
We have external support that had completed the work. They stated they updated the memory allocation for Manager Server on IFS MWS:
They increased it to;
Lower limit 0.4gb
Upper Limit 0.6gb
We currently are licensed for 200 users and use 135. Realistically we peak at around 40 active connections maximum currently.
When I check the middleware server admin I can see both our initial memory size and maximum is similar to yours.
So 0.4 GB is really only about 400 MB are you should it was 0.4 and not 4? 0.4 GB is not enough to handle the start up of the application server.
Also did they do this increase on the OS server instance itself or just in the application server?
What are these settings set to?
Having too many Free Connections or having too long a time out can cause the connection pool to fill up with connections that should have been recycled.
We have external support that are dealing with this and this is the information they sent across:
The lower limit it 4 GB and the upper limit is 6 GB for both Managed Server memory allocation for IFS MWS
Our OS memory has 32gb allocated to it. Normally 20gb is being used by Java over 5 different Processes.
The settings you’ve requested are:
We currently don’t have a huge IFS user count. Maybe 40 active connections at peak so I can’t understand why we’re having memory related problems unless there’s a memory leak or the likes?