Question to core-processes

Question

We currently have following configuration set regarding processes:
<Process>
    <Scope>app</Scope>
    <Distribute numprocs="11">insts</Distribute>
    <LifeTime>
      <RollOver></RollOver>
      <MaxCores></MaxCores>
    </LifeTime>
  <MaxFailures>5</MaxFailures>
                    <RecoveryTime>5</RecoveryTime>
</Process>
We are running non-stateless-applications so we can't use the roll-over-configuration.
This leads to instances running for over 30 days (with several connects and disconnects) resulting in core-processes, that don't seem to work properly after about 30 days. They still respond but use a high amount of virtual memory and seem to have utilization-related problems.
What we could do is to manually stop an instance via the application itself, if a certain state is matching. But we would have to immediately restart the instance.
With the configuration above, is it possible, that when i restart an instance via the Flash Media Administration Console it is assigned to the same core-process or will FMS assign the instance to another core-process?
Is there a way to ensure, that the instance gets assigned to a different core-process, so the old core process will be stopped?
Any help is appreciated.
Best regards
Suha

Answer 1

Well i guess so, too, but it would help to be sure.
Just to make sure: We don't want to run the same instance in the same old core-process! I want to make sure, that if i terminate an instance via code on a certain state it gets assigned to fresh new core-process.
I once tried to simply restart an instance via Flash Media Admin Console, that was running a long time and did not write any more instance-log-files. But after restart it still did not write any more log-data. So my guess was, that the restarted instance was not assigned to a new core-process but to the same old core-process, that was for whatever reason not able to write any more log-data.
Other instances of the same application running on the same machine still did write log-data.
If i did not get the documentation wrong, our current configuration would tell FMS to assign all instances of one application to up to eleven core-processes. How can we make sure, that an old, long-running core-process gets terminated? I guess, that all still running instances in this process have to be stopped.
But we can't do it that way, we could only try to restart instances, which only makes sense, when it is for sure, that a restarted application-instance will not be assigned to the same old process neither newly started instances will. Is there a way, without using the rollover-configuratiuon to achieve this?