Re: Memory leak?

[Jeffrey Ross]

Nathan Grennan wrote:
> Jeffrey Ross wrote:
> > > I believe I have a memory leak someplace on the system,  over several 
> > > days I watch the used memory slowly climb and eventually the system 
> > > starts swapping, the application load has not changed.  I have tried 
> > > stopping processes and restarting them in the hopes that the memory 
> > > would be returned to the system but no such luck.
> > > Kernel 2.4.16-1
> > > How can I track down the source of the memory leak?
> > >
> > > according to "top" spamd (spamassassin) is the largest consumer of 
> > > memory, however it seems to stay pretty stable compaired to how fast 
> > > memory usage is climbing
> > >
> > > TIA, Jeff
>
> I saw a similar problem with kernel-2.6.15-1.1824_FC4 on one server. I 
> tracked it back to slab debugging.
>
> https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=177593

First, my mistake, the kernel is 2.6.14-1.1656_FC4 not 2.4.16-1

is slab debugging turned on in this kernel?  and if so how do I disable 
it without rebuilding the kernel?  can I write something to /proc/???

Part of my problem is this is a memory challenged system and due to 
hardware constraints I can't up the memory without changing the 
motherboard.  The only saving grace is I am not running a gui interface 
on the machine.

thanks

Jeff

Memory leak?

[Nathan Grennan]

> First, my mistake, the kernel is 2.6.14-1.1656_FC4 not 2.4.16-1
>
> is slab debugging turned on in this kernel?  and if so how do I 
> disable it without rebuilding the kernel?  can I write something to 
> /proc/???

No, 2.6.14-1.1656_FC4 should not have the problem. That is actually 
the kernel the server that was having the problem is currently running. 
The slab debugging is only on in test/developments kernels as far as I 
know, and  2.6.14-1.1656_FC4 is a normal kernel. But your problem still 
sounds a lot like mine. I suggest examining /proc/meminfo and 
/proc/vmstat for clues. Also compare the results from a working kernel 
and a problem kernel.