Bug#554003: cpu plugin has too strict limits

Peter Palfrader · External Since: Nov 10, 2004 Posts: 281

Package: munin-node
Version: 1.2.5-1
Severity: normal
Tags: patch

Hi,

the cpu munin plugin configures its DERIVE fields with min and max.
Minimum is set to 0, and maximum is set to 100*numcpus. It gets its
values from /proc/stat.

The problem is that if your system is mostly in one state (e.g. idle)
then the derived values over time sometimes are slightly larger than the
configured max. For instance it might be 100.12 instead of 100 for one
second of idling.

Munin-update/rrd then throws away these data points.

A nice example of what happens can be found at
http://asteria.noreply.org/~weasel/cpu-week.png

An easy fix is to slightly raise the configured max:

--- /usr/share/munin/plugins/cpu 2009-03-13 14:10:06.000000000 +0100
+++ /usr/local/bin/munin-cpu 2009-11-02 11:48:21.000000000 +0100
@@ -47,6 +47,7 @@
SYSWARNING=`expr $PERCENT '*' 30 / 100`
SYSCRITICAL=`expr $PERCENT '*' 50 / 100`
USRWARNING=`expr $PERCENT '*' 80 / 100`
+ MAX=$(($MAX * 120 / 100))
echo 'graph_title CPU usage'
echo "graph_order system user nice idle" $extinfo
echo "graph_args --base 1000 -r --lower-limit 0 --upper-limit $graphlimit"

If you agree with the fix then seeing this in a stable point release
would be nice. It makes the cpu graphs pretty pointless at the moment.

Cheers,
weasel

--
To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST RemoveThis @lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster RemoveThis @lists.debian.org

Peter Palfrader · External Since: Nov 10, 2004 Posts: 281

On Tue, 03 Nov 2009, Tom Feiner wrote:

> forwarded 554003 http://munin.projects.linpro.no/ticket/736
> tags 554003 - patch
> quit

> The cpu plugin has been rewritten since version 1.2.5 which you are using, so
> your patch will probably not apply.

The patch tag still applies to the stable version.

--
| .''`. ** Debian GNU/Linux **
Peter Palfrader | : :' : The universal
http://www.palfrader.org/ | `. `' Operating System
| `- http://www.debian.org/

--
To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST.DeleteThis@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster.DeleteThis@lists.debian.org

Tom Feiner · External Since: Aug 24, 2009 Posts: 5

On Tue, Nov 3, 2009 at 7:44 PM, Peter Palfrader <weasel DeleteThis @debian.org> wrote:
> On Tue, 03 Nov 2009, Tom Feiner wrote:
>
>> forwarded 554003 http://munin.projects.linpro.no/ticket/736
>> tags 554003 - patch
>> quit
>
>> The cpu plugin has been rewritten since version 1.2.5 which you are using, so
>> your patch will probably not apply.
>
> The patch tag still applies to the stable version.

Right! I see you've re-tagged it. Thanks!

Tom

--
To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST DeleteThis @lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster DeleteThis @lists.debian.org

Peter Palfrader · External Since: Nov 10, 2004 Posts: 281

On Tue, 03 Nov 2009, Tom Feiner wrote:

> On Tue, Nov 3, 2009 at 7:44 PM, Peter Palfrader <weasel.TakeThisOut@debian.org> wrote:
> > On Tue, 03 Nov 2009, Tom Feiner wrote:
> >
> >> forwarded 554003 http://munin.projects.linpro.no/ticket/736
> >> tags 554003 - patch
> >> quit
> >
> >> The cpu plugin has been rewritten since version 1.2.5 which you are using, so
> >> your patch will probably not apply.
> >
> > The patch tag still applies to the stable version.
>
> Right! I see you've re-tagged it. Thanks!

I have not.

--
| .''`. ** Debian GNU/Linux **
Peter Palfrader | : :' : The universal
http://www.palfrader.org/ | `. `' Operating System
| `- http://www.debian.org/

--
To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST.TakeThisOut@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster.TakeThisOut@lists.debian.org

Tom Feiner · External Since: Aug 24, 2009 Posts: 5

tags 554003 + patch
quit

Oops, browser cache Smile

Tom

--
To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST RemoveThis @lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster RemoveThis @lists.debian.org

Peter Palfrader · External Since: Nov 10, 2004 Posts: 281

On Mon, 02 Nov 2009, Peter Palfrader wrote:

> An easy fix is to slightly raise the configured max:
>
> --- /usr/share/munin/plugins/cpu 2009-03-13 14:10:06.000000000 +0100
> +++ /usr/local/bin/munin-cpu 2009-11-02 11:48:21.000000000 +0100
> @@ -47,6 +47,7 @@
> SYSWARNING=`expr $PERCENT '*' 30 / 100`
> SYSCRITICAL=`expr $PERCENT '*' 50 / 100`
> USRWARNING=`expr $PERCENT '*' 80 / 100`
> + MAX=$(($MAX * 120 / 100))
> echo 'graph_title CPU usage'
> echo "graph_order system user nice idle" $extinfo
> echo "graph_args --base 1000 -r --lower-limit 0 --upper-limit $graphlimit"
>
>
> If you agree with the fix then seeing this in a stable point release
> would be nice. It makes the cpu graphs pretty pointless at the moment.

I have not heard anything from you munin folks about a possible fix of
this bug in stable. Comments?

weasel
--
| .''`. ** Debian GNU/Linux **
Peter Palfrader | : :' : The universal
http://www.palfrader.org/ | `. `' Operating System
| `- http://www.debian.org/

--
To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST.RemoveThis@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster.RemoveThis@lists.debian.org

Peter Palfrader · External Since: Nov 10, 2004 Posts: 281

On Mon, 16 Nov 2009, Tom Feiner wrote:

> So we should probably continue this discussion in the upstream bug.

Or we could just fix it.

> However, I've seen that so far upstream developers haven't responded to
> this bug. Maybe some more details on when exactly this is happening
> might (how to reproduce exactly) might help.

I have lined out whent it happens. And it happens lots and often.

> Another thing that I notice, is that according to this bug, you're
> running munin version 1.2.5-1 which is the one released for etch. The
> cpu plugin was enhanced in munin version 1.2.6, which is packaged for
> lenny.

Bad copy/paste. It applies to lenny, not etch (bug report accordingly
modified).
--
| .''`. ** Debian GNU/Linux **
Peter Palfrader | : :' : The universal
http://www.palfrader.org/ | `. `' Operating System
| `- http://www.debian.org/

--
To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST DeleteThis @lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster DeleteThis @lists.debian.org

Peter Palfrader · External Since: Nov 10, 2004 Posts: 281

On Mon, 16 Nov 2009, Tom Feiner wrote:

> Peter Palfrader wrote:
> > On Mon, 16 Nov 2009, Tom Feiner wrote:
> >
> > Bad copy/paste. It applies to lenny, not etch (bug report accordingly
> > modified).
>
> Thanks for updating the found version.
>
> After digging a bit more, this looks like the same issue as debian bug [1] ,
> which is fixed in testing. They used a similar fix as you proposed, only
> instead of increasing the MAX field slightly, they removed it completely.
>
> This is also the same solution offered by upstream [2], only backported to
> munin 1.2.6.
>
> Can you test if munin from testing fixes your problem?

http://patch-tracker.debian.org/patch/series/view/munin/1.2.6-17/610-p...in-cpu-

If that's the patch you are referring to then it very likely fixes my
bug. It's a more brutal version of my patch, both will do.

--
| .''`. ** Debian GNU/Linux **
Peter Palfrader | : :' : The universal
http://www.palfrader.org/ | `. `' Operating System
| `- http://www.debian.org/

--
To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST.RemoveThis@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster.RemoveThis@lists.debian.org