[CentOS] Ignoring /run/user/X

Cameron Smith cameron at networkredux.com
Wed Apr 12 15:43:02 UTC 2017


Thanks Tony.

Looks like it's possibly handled in this file:
https://sourceforge.net/p/net-snmp/code/ci/master/tree/agent/mibgroup/host/hr_disk.c

I will dig :)

Cameron

On Wed, Apr 12, 2017 at 5:26 AM, Tony Mountifield <tony at softins.co.uk>
wrote:

> In article <CABLobCMg4irXK7Cw6xYhz9g6_0Jug9_Lq33zVyHdQE-JAxoPvw@
> mail.gmail.com>,
> Cameron Smith <cameron at networkredux.com> wrote:
> > We are running into an issue relating to snmpd and the temporary
> partitions
> > created in /run/user/ so any insight by someone with magical
> > net-snmp skills would be much appreciated.
> >
> > Our monitoring app walks all our servers.
> > We modify /etc/snmp/snmpd.conf on all our servers to just have one line:
> > rocommunity ourcommnuityname monitor.ing.app.ip
> >
> > This has worked just fine for almost 10 years.
> >
> > Since the release of CentOS 7 we are getting alerts for partitions not
> > being found during walks and these are the temporary partitions that are
> > ephemeral while a user is logged in:
> > /run/user/0
> > /run/user/65000
> >
> > Seems if the partition is there when the monitor is set or is added while
> > the monitor is active the monitor will keep looking for it and these are
> > meant to go away.
> >
> > We thought "OK we can just ignore those" so we added a line to
> > /etc/snmpd.conf :
> > ignoredisk /run/user/*
> >
> > but that has not helped :(
> >
> > Does anybody have a recommendation on how we can stop those partitions
> from
> > being seen on walks so we can stop being alerted about partitions for
> which
> > we are not interested in monitoring their available space?
>
> I had a quick play with it on a C7 VM, and found the same as you have.
>
> It would appear that ignoredisk only allows you to specify device names
> (such as /dev/sda1 or /dev/cciss/*), and not mount points. For /run and
> /run/user/*, they are not mounted on devices but on tmpfs.
>
> I tried "ignoredisk tmpfs" to see if that would work, but it didn't appear
> to.
>
> It also doesn't help that the SNMP output for the mount table doesn't seem
> to include a column for the device that was mounted, only for the mount
> point.
> And all the mount points are listed as type "hrFSOther", so you can't tell
> the difference between real disks, tmpfs, and so on.
>
> You probably need to get the SRPM for net-snmp and have a look at the area
> of code that process "ignoredisk".
>
> Cheers
> Tony
> --
> Tony Mountifield
> Work: tony at softins.co.uk - http://www.softins.co.uk
> Play: tony at mountifield.org - http://tony.mountifield.org
> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> https://lists.centos.org/mailman/listinfo/centos
>



More information about the CentOS mailing list