[Linux-HA] Re: Memory Leak by crmd in 2.0.7
Andrew Beekhof
beekhof at gmail.com
Thu Feb 8 04:52:39 MST 2007
On 2/8/07, kisalay <kisalay at gmail.com> wrote:
> Thanks for the reply.
>
>
> I observer that the crmd seems to leak memory till it fills up its heap and
> then it aborts.
> Will this destabilize the system ?
> Is this abort from crmd and a subsequent core only specific to my setup ? or
> is it reproducable elsewhere as well ?
It is reproducible elsewhere.
I'm having a lot of success with valgrind and hope to have the crm
leak free by the end of today or tomorrow. At that time I'll post a
comprehensive report and a patch against 2.0.8.
>
> I am planning to use HA for my company's cluster product. Please give some
> insight in this.
>
> Thanks
> Kisalay
>
>
>
> ------------------------------
> >
> > Message: 2
> > Date: Wed, 7 Feb 2007 18:51:55 +0530
> > From: kisalay <kisalay at gmail.com>
> > Subject: [Linux-HA] Memory Leak by crmd in 2.0.7
> > To: linux-ha at lists.linux-ha.org
> > Message-ID:
> > <d1537de90702070521k73789410wc33c7e2b23e61dad at mail.gmail.com>
> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> >
> > Hi,
> >
> > I have a 2.0.7 HA installed. The crmd process seems to continuosly leak
> > memory.
> > I am posting below the memory occupied by crmd:
> >
> > I started these observations after around 12 hours of crmd uptime.
> > ElapsedTime(secs), VmSize, VmRSS
> > 0, 1010872, 1008844
> > 600, 1019172, 1017252
> > 1200, 1027736, 1025704
> > 1800, 1036160, 1034176
> > 2400, 1044580, 1042624
> > 3000, 1052992, 1050996
> >
> > the ouput from /proc is:
> >
> > 0931e000-56dac000 rw-p 0931e000 00:00 0 [heap]
> > b7f47000-b7f4c000 rw-p b7f47000 00:00 0
> > b7f53000-b7f55000 rw-p b7f53000 00:00 0
> > bf907000-bf98c000 rw-p bf907000 00:00 0 [stack]
> >
> >
> > after another 600 seconds:
> >
> > 0931e000-575c5000 rw-p 0931e000 00:00 0 [heap]
> > b7f47000-b7f4c000 rw-p b7f47000 00:00 0
> > b7f53000-b7f55000 rw-p b7f53000 00:00 0
> > bf907000-bf98c000 rw-p bf907000 00:00 0 [stack]
> >
> >
> > Is this memory leak inherent in crmd in 2.0.7 release. If so, is it fixed
> > in
> > the latest 2.0.8 release.
> >
> > I will vey much appreciate any help in this regard.
> >
> > Thanks,
> > Kisalay
> >
> >
> > ------------------------------
> >
> > Message: 3
> > Date: Wed, 7 Feb 2007 14:40:38 +0100
> > From: "Andrew Beekhof" <beekhof at gmail.com>
> > Subject: Re: [Linux-HA] Memory Leak by crmd in 2.0.7
> > To: "General Linux-HA mailing list" <linux-ha at lists.linux-ha.org>
> > Message-ID:
> > <26ef5e70702070540k5418248hcdd5964d1e75ccd3 at mail.gmail.com>
> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> >
> > On 2/7/07, kisalay <kisalay at gmail.com> wrote:
> > > Hi,
> > >
> > > I have a 2.0.7 HA installed. The crmd process seems to continuosly leak
> > > memory.
> > > I am posting below the memory occupied by crmd:
> > >
> > > I started these observations after around 12 hours of crmd uptime.
> > > ElapsedTime(secs), VmSize, VmRSS
> > > 0, 1010872, 1008844
> > > 600, 1019172, 1017252
> > > 1200, 1027736, 1025704
> > > 1800, 1036160, 1034176
> > > 2400, 1044580, 1042624
> > > 3000, 1052992, 1050996
> > >
> > > the ouput from /proc is:
> > >
> > > 0931e000-56dac000 rw-p 0931e000 00:00 0 [heap]
> > > b7f47000-b7f4c000 rw-p b7f47000 00:00 0
> > > b7f53000-b7f55000 rw-p b7f53000 00:00 0
> > > bf907000-bf98c000 rw-p bf907000 00:00 0 [stack]
> > >
> > >
> > > after another 600 seconds:
> > >
> > > 0931e000-575c5000 rw-p 0931e000 00:00 0 [heap]
> > > b7f47000-b7f4c000 rw-p b7f47000 00:00 0
> > > b7f53000-b7f55000 rw-p b7f53000 00:00 0
> > > bf907000-bf98c000 rw-p bf907000 00:00 0 [stack]
> > >
> > >
> > > Is this memory leak inherent in crmd in 2.0.7 release. If so, is it
> > fixed in
> > > the latest 2.0.8 release.
> > >
> > > I will vey much appreciate any help in this regard.
> >
> > in the last day or so I've begun working on tracking down all the
> > memory leaks in the CRM (including the crmd). there are some noisy
> > leaks in Alan's messaging layer that make the task a little harder but
> > I have already been able to make some progress.
> >
> > The most effective method, if you'd like to help, is to define
> > HA_TRACK_MALLOC in cl_malloc.h and add:
> >
> > #ifdef HA_MALLOC_TRACK
> > cl_malloc_dump_allocated(LOG_ERR, FALSE);
> > #endif
> >
> > just before the various processes exit. this will indicate who
> > created the remaining unfree'd memory.
> >
> > running against the latest from http://hg.linux-ha.org/dev will also
> > make your (and my) life easier.
> >
> >
> > _______________________________________________
> > Linux-HA mailing list
> > Linux-HA at lists.linux-ha.org
> > http://lists.linux-ha.org/mailman/listinfo/linux-ha
> > See also: http://linux-ha.org/ReportingProblems
> >
> > End of Linux-HA Digest, Vol 39, Issue 22
> > ****************************************
> >
> _______________________________________________
> Linux-HA mailing list
> Linux-HA at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>
More information about the Linux-HA
mailing list