Nodemgr process running away with memory

Question

We have 3 different applications that this has happened to. All of a
sudden, the nodemgr for the application environment will start consuming
memory on the Unix Solars 2.6 box. The nodemgr process gets up to
2 gig of virtual memory usage and seems to 'die', not to mention sometimes
consuming all the available memory on the box, locking up the box and
requiring a box reboot. I'm getting paged when a process starts consuming
memory so at least a box reboot has been avoided lately.
One application is using 3G2, two 3f2 and one is a web sdk application.
I suspect that there is a bug in Forte causing this to happen. Maybe a
bug with the way Forte is not handling something from the version of Solaris
we're using.
I've been working with a Forte tech rep. and they have said they haven't
seen any customers with the problem and don't have a reason for it.
Has anyone else seen this happen and can you please shed some light
on it?
Thanks,
Peggy Adrian
[email protected]
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>

Answer 1

Peggy:
I have DEFINITELY seen this problem and strongly believe that it is related to
3G2 running on Solaris 2.6. As you know, 3G2 is not supported on Solaris 2.6.
We upgraded to 3J1 and Solaris 2.6 in late November, and have never seen this
problem since. However, we did see this problem a number of times when trying
to run 3G2 on Solaris 2.6.
By the way, we did find that if we killed the nodemgr and then restarted it
VERY quickly, it would rebind properly and all would be well (except for
having to start all the router partitions again.)
-Martin
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>