On Wed, May 09, 2007 at 10:24:49PM +0200, Rik Theys wrote:
> On a busy NFS server, rpc.mountd starts to slowly eat a _lot_ of memory. I
> usually restart
> it when the amount of memory reaches 1.6Gb (which is approx. 10% of the
> system memory).
> This is after approx. 2 weeks of uptime.
>
> This bug looks a lot like the bug described here:
> http://thread.gmane.org/gmane.linux.nfs/13479/focus=13480
Hm, I'm pretty sure that leak is specific to 1.1.0, at least the major part of
it (the large leak was in libblkid1, which is not used in 1.0.x). There's a
different, tiny leak, in mountd itself, though; if you look further down the
thread, you'll see this patch which I wrote to fix the 1.1.0 issue. Could you
please try it under 1.0.12 and see if it fixes your problem?
--- a/support/export/client.c
+++ b/support/export/client.c
<at> <at> -329,6 +329,7 <at> <at> add_name(char *old, char *add)
strcat(new, ",");
strcat(new, cp);
}
+ free(old);
return new;
}
> If this bug is already fixed in new upstream versions (didn't check), do you
> think it's
> possible to backport the fix to stable and ask for it's inclusion in a stable
> point release?
I believe it should be applicable, but I'm not sure if this is the bug you're
actually seeing. Do you have tons of mounts and umounts?
/* Steinar */
--
Homepage: http://www.sesse.net/
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
