Checkmk Werk 7444: Rename metric name in Filesystem checks from mount point to fs_used

Óscar Nájera oscar.najera at
Mon Jul 15 14:46:40 CEST 2019

ID:          7444
Title:       Rename metric name in Filesystem checks from mount point to fs_used
Component:   Core & setup
Level:       3
Class:       Bug fix
Version:     1.7.0i1

Filesystem Nagios check plug-in stored the filesystem usage of every mount
points in the RRD databases under the path of each mount point. Checkmk
assigned to every mount point a service, still it kept naming the
filesystem usage with the mount point path for compatibility. Our graphing
system learned to deal with this unpractical convention, however new
features in Checkmk performing bulk access to data can't.


<li>All sites created with Checkm 1.6.0 onwards will automatically use the
new naming convention and don't need to be migrated</li>

<li> The migration is not mandatory, your Checkmk instance will continue
working under the legacy mode until you migrate.</li>

<li>If you use CRE, you don't need to de the migration</li>

<li>If you don't plan to do bulk access to data as required by the Historic
Data Views for filesystems (see Werk #7445), you don't need to migrate your
RRD databases.</li>

<li>If you want to use the Historic Data Views for Capacity Management on
Filesystems, you need to do migration.</li>


The incorporation of Historic Data Views, for Capacity Management,
requieres querying the filesystem usage of all host in a given time
window. It becomes prohibitive expensive to do such request since first
every service is queried for its mount point name and then every RRD is
individually queried again for the mount point name data.

In this werk we provide a migration script that would edit all meta data
files describing the RRDs that receive output from all our supported
filesystem checks. It will also set a flag in your config so that
Filesystem check plug-ins start delivering usage perfomance data under the
static name of fs_used.

You are advised to do a backup of your RRD files and journal files before
performing the migration and your Checkmk instance needs to be stopped
during the migration.

Migration is one way and needs to be performed once per site any rollback
has to be done from your own backup.

More information about the checkmk-werks-lvl2 mailing list