[Pkg-nagios-devel] Bug#395389: Bug#395389: nagios2: complains about disk space in an uncomprehensible way

[Filip Van Raemdonck]

On Thu, Jun 05, 2008 at 11:45:43AM +0200, Jan Wagner wrote:
> On Saturday 28 October 2006 11:56, Marc Haber wrote:
> > On Thu, Oct 26, 2006 at 08:48:36PM +0200, Marc Haber wrote:
> > > Formatting of plugin output is a plugin issue. Do you want me to
> > > reassign or clone to nagios-plugins?
> >
> > After investigating, this has nothing to do with nagios2 at all.
> > Reassigning to nagios-plugins-basic.
> >
> > check_disk has an option --errors-only, and you can ignore everything
> > in the output after the |. That might help in pinning down what your
> > check_disk thinks is wrong.
> >
> > Please show the actual output of
> > /usr/lib/nagios/plugins/check_disk -w 20 -c 10 --errors-only.
> 
> Could you please verify if the problem is present on unstable and if so, 
> provide the output of the command Marc was asking for?

Cannot, unfortunately, I do not have access to the system (or any other
Nagios system) anymore.

Let's review the issues in the original report:

1) Output is a huge line, which is cut off;
2) in the output visible there's nothing present which would warrant a
warning or error;
3) but Nagios is showing a critical state for disk space;
4) according to follow-up message from Marc, the larger part of the output
is just informational and it's the fact that there is a "DISK CRITICAL"
message at the front which is really important.

Unfortunately I failed to mention whether there really was a partition
with little free space remaining, but given that I didn't and seeing how I
phrased the report I assume there was.

So, to clarify the report:
When nagios complains that disk space is critical it should tell me which
partition that is, not let me figure it out from a one-line dump of df(1)
output. Which isn't even complete.
As a user, I don't care whether it's really the plugin's responsibility to
take care of the output. Though if it is, I agree with the reassigment
that Marc did.
And I don't think the output of running that check_disk plugin manually
really matters, unless the --errors-only switch would cause all the
"compliant" partitions to be removed (but still list the lacking-space
one) in which case the effect of that switch should be applied by default,
on the check_disk plugin.

(The best way to reproduce the issue, or see if it's still present, is to
add a bunch of loopback mounted images -- could be dummies -- until the
regular check_disk output gets too long. And make sure at least one
partition causes a warning or error, off course)


KR,

Filip

-- 
My writings: http://blog.sysfs.net/