Fwd: strange memory behavior with oM

---------- Forwarded message ----------
From: jafarim <[email protected]>
Date: May 31, 2006 1:04 PM
Subject: Re: [openMosix-general] strange memory behavior with oM
To: moshe-ay74M1d3r6RWk0Htik3J/[email protected]

I have a written a small program that acts as a scheduler. It gets a list of
job to be run and then selects a small portion of these jobs and runs them
on the cluster. The jobs are of two kind. Type-1 jobs are generators that
generate a file which type-2 jobs then consume. I use Qt 4.1.2 for wirting
the scheduler and there may be memory leaks as I am not a very skilled
programmer. But if you remeber I said that even when I stop this scheduler
the memory is not freed and this is only happening on remote nodes and not
the home node where I start the scheduler. The memory is seemingly allocated
to kernel as the total memory of all processes (as shown in top) is very low
and insignificant.


On 5/31/06, Moshe Bar <moshe-ay74M1d3r6RWk0Htik3J/[email protected]> wrote:
>
> I was not aware of memory leaks for 2.4.26. What kind of app are you
> running? What libraries does it use?
>
> Moshe
>
> -----Original Message-----
> From: openmosix-general-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> [mailto:openmosix-general-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org] On Behalf Of
> jafarim
> Sent: Tuesday, May 30, 2006 9:20 AM
> To: openmosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> Subject: [openMosix-general] strange memory behavior with oM
>
> Hi
> I encountered a very strange problem with openMosix recently. I have a
> 3-node cluster with dual AMD Opteron CPUs and 1 GB of memory each. My
> cluster is based on Fedora Ccore 1 on which I have installed openMosix
> rpms
> (2.4.26). I have a small manager which queus jobs and starts them on
> random
> nodes by using mosrun.
> The problem is that when there is a large load on the cluster, the memory
> used by kernel grows gradually (even to 512 MB after hours of load).
> Restarting my manager program or even openMosix daemon does not help and
> the
> only treatment is to reboot the nodes. I was firstly suspicious about
> memory
> leak in my manager program but the most strange observation is that this
> memory growth does not happen on the node where my manager program resides
>
> (and starts other jobs) and that node's memory is quite normal. I verified
> this observation by moving my manager program to some other nodes in the
> cluster.
> Any ideas?
>
>
>
> -------------------------------------------------------
> All the advantages of Linux Managed Hosting--Without the Cost and Risk!
> Fully trained technicians. The highest number of Red Hat certifications in
> the hosting industry. Fanatical Support. Click to learn more
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=107521&bid=248729&dat=121642
> _______________________________________________
> openMosix-general mailing list
> openMosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> https://lists.sourceforge.net/lists/listinfo/openmosix-general
>