Memory Issue - huge RAM consumed even no users logged in

[weblike]

Hi,

I see that the memory on my server is kept to 6<->7GB even no user is logged in. In the past I have read somewhere that it concerns a table form DB which gets huge.
Can someone point a suggestion on this?

Thank you,

[tford]

There is a table called Execution Contexts or some variation of that. You can search the forums about it. Not sure if it impacts memory or now.

[BenHayat]

There is a table called Execution Contexts or some variation of that. You can search the forums about it. Not sure if it impacts memory or now.

I had told George a while back regarding this table and clear it but he never followed up with me.

Rod, got to clear that table in one of threads and saw great improvements in memory.

[weblike]

There is a table called Execution Contexts or some variation of that. You can search the forums about it. Not sure if it impacts memory or now.

I had told George a while back regarding this table and clear it but he never followed up with me.

Rod, got to clear that table in one of threads and saw great improvements in memory.

No results on my side.

[aware_support]

Does the memory keep growing or it just uses this amount when you start the app? If so, there must be something huge in your app that gets loaded.

[weblike]

Does the memory keep growing or it just uses this amount when you start the app? If so, there must be something huge in your app that gets loaded.

When I start the app, everything is ok.
In the morning when users keeps login the memory increases to 6<->7GB of RAM (users are calling that the app is freezing, and I have to restart the app).
Strange thing that in the end-of-the-day the memory doesn't go down, so in the morning I have to restart the app to go fresh again.

Today I have made a remote session with a user and the app is slow when user search in drop-downs (For example I have a database of customers-> when I click on drop-down to search in archivefor a customer).

Any suggestion would be greatly appreciated !!!!
Thank you

[aware_support]

There must be something in your app that causes this. You need to monitor what the users are doing and what's happening with memory - this will give you a clue which operation is causing the problem.

[weblike]

There must be something in your app that causes this. You need to monitor what the users are doing and what's happening with memory - this will give you a clue which operation is causing the problem.

Ok. But how do you explain that after job hours when everyone leave the app, the RAM is not coming back to normal. What makes it to stay high?Something got stuck?

[codrin]

At times we've found similar behaviour and tried at length to find one specific reason by monitoring the system usage over extended periods of time without any conclusive outcome. In the end, we have decided to use a small utility which constantly monitors the memory allocation and regularly cleans and resets it. We have now used this approach for all our windows servers for years with 100% success.
http://www.majorgeeks.com/mg/getmirror/cleanmem,1.html

[tford]

http://www.majorgeeks.com/mg/getmirror/cleanmem,1.html

Can you tell us a bit about how this works in the AwareIM environment?

[BenHayat]

At times we've found similar behaviour and tried at length to find one specific reason by monitoring the system usage over extended periods of time without any conclusive outcome. In the end, we have decided to use a small utility which constantly monitors the memory allocation and regularly cleans and resets it. We have now used this approach for all our windows servers for years with 100% success.
http://www.majorgeeks.com/mg/getmirror/cleanmem,1.html

Cordin, when you say to clean up memory, do you mean, java objects that were created by aware and were never "Disposed"? Because this problem seems like there are certain types of objects created and as long as Aware is still hanging on to them (not disposed explicitly) Java runtime will keep those objects.
Now how does this tool know which objects to be disposed? Do you run this tool while Aware is running (which assume yes).

[codrin]

It works on a double front, not only with AwareIM but any other app installed on the server. It will "clean" the allocated memory down to only what values are required by the system at that point in time. I am not entirely sure what exact process takes place in the windows environment, but I can assure you that it does not affect AwareIM user sessions or performance. It is almost as if Java's native garbage collection is invoked and executed. You can have it on a timer to execute every so many minutes ( we have ours set to 15 minutes ) and also to kick in when the max allocation of memory reaches a certain level ( ours is set to 85% ). Another welcome side effect of using it was the fact that we could increase the number of accounts on a given server by at least 20% with the same hardware configuration.

[BenHayat]

It works on a double front, not only with AwareIM but any other app installed on the server. It will "clean" the allocated memory down to only what values are required by the system at that point in time. I am not entirely sure what exact process takes place in the windows environment, but I can assure you that it does not affect AwareIM user sessions or performance. It is almost as if Java's native garbage collection is invoked and executed. You can have it on a timer to execute every so many minutes ( we have ours set to 15 minutes ) and also to kick in when the max allocation of memory reaches a certain level ( ours is set to 85% ). Another welcome side effect of using it was the fact that we could increase the number of accounts on a given server by at least 20% with the same hardware configuration.

Very interesting. Basically puts GC on steroids.
Thank you Cordin!!!

[aware_support]

Aware IM has been extensively tested for memory leaks. It is unlikely that this is the problem. Of course, we cannot rule out the leaks completely - there might be some rarely used operation that causes a leak. But it is still unlikely.

The fact that the memory is not released is not an indication of a leak, but could be an indication of a problem in your application - that's why you need to monitor it. You have to understand that Aware IM provides a framework but a lot depends on the application itself - this relates to both memory usage and performance.

Now Codrin's utility performs some magic at the operating system level - as a result it improves the performance of the Java garbage collector. The fact that it successfully cleans up memory indicates that Aware IM itself does not have a problem (otherwise, memory wouldn't have been released), but the operating system may not work as efficiently as it could. So the utility can be very useful as Codrin says.

It is possible that the utility will solve George's problem if it is indeed caused by the operating system. But it is also possible that it won't - if the problem is caused by the application itself.

[weblike]

At times we've found similar behaviour and tried at length to find one specific reason by monitoring the system usage over extended periods of time without any conclusive outcome. In the end, we have decided to use a small utility which constantly monitors the memory allocation and regularly cleans and resets it. We have now used this approach for all our windows servers for years with 100% success.
http://www.majorgeeks.com/mg/getmirror/cleanmem,1.html

Hi Codrin and many thanks for your solution.

The cleanmem app indeed releases memory a lot, but right after the scheduler runs is about aprox 15 seconds time when the app is reloading, after that is running ok.

Regarding what support says, I will follow your optimization video and try to review my rules.

Thank you all.