Dockerized version of Observium with support for external database, both communitiy and professional editions (via arbritrary svn source), packages to allow native LDAP auth, and easy plugin support by exposing htmldir as a volume.
At Yelp we use this to provide portability to our NMS. With a flexible image, it's easy to manage 2+ instances with Puppet or [insert way to configure/schedule containers here]. Good use cases would be instances for corp and prod, managed services and internal, or just a single one that's more predictable on any box.
Examples assume the following directory layout:
$ tree $PWD
/opt/observium/volumes
├── config
│ └── config.php
├── html
├── logs
└── rrd
Linking volumes and using Observium CE:
$ docker run -d \
--name observium \
-p 8000:8000 \
-v /opt/observium/volumes/config:/config \
-v /opt/observium/volumes/html:/opt/observium/html \
-v /opt/observium/volumes/logs:/opt/observium/logs \
-v /opt/observium/volumes/rrd:/opt/observium/rrd \
yelp/observium
Using Observium PE (or another SVN source):
$ docker run -d \
--name observium \
-p 8000:8000 \
-v /opt/observium/volumes/config:/config \
-v /opt/observium/volumes/html:/opt/observium/html \
-v /opt/observium/volumes/logs:/opt/observium/logs \
-v /opt/observium/volumes/rrd:/opt/observium/rrd \
-e USE_SVN=true \
-e [email protected] \
-e SVN_PASS=dfed555743854a475345ae01a7668acc \
-e SVN_REPO=http://svn.observium.org/svn/observium/trunk \
yelp/observium
The following volumes are set in the container:
Volume | Purpose |
---|---|
/config |
config.php should go here |
/opt/observium/html |
This allows you to add HTML plugins if you wish (such as weathermap!) |
/opt/observium/logs |
HTTP error/access, observium.log and update-errors.log |
/opt/observium/rrd |
Mount this where you have a ample space and back up! |
/var/run/mysqld/mysqld.sock |
If you're running MySQL on the local Docker host, make use of this volume and set the SQL host to localhost in config.php |
If you need to sub-in random, single files just add an extra -v
argument to
your run (or however you're starting the container. Puppet anyone?) to the
path. We do this for /etc/php5/apache2/php.ini
Only TCP/8000 is exposed.
Variable | Default Value |
---|---|
WORKERS |
2 |
USE_WEATHERMAP |
false |
CUSTOM_PHP_INI |
false |
HOUSEKEEPING_ARGS |
'-yet' |
USE_SVN |
false |
SVN_USER |
N/A |
SVN_PASS |
N/A |
SVN_REPO |
N/A |
If you don't have a subscription but run your own managed copy, feel free to sub-in your own repo.
Depending on your number of devices, the default 2 workers likely isn't enough
based on a polling period every 5 minutes. Just pass WORKERS=32
, for
example, if your total core count is 32 and you want to use them all.
As in the list above, setting USE_WEATHERMAP
will install network-weathermap
under the htmldir if not there already and schedule map-poller.php in cron.
To modify configuration, volume mount observium's htmldir to access weathermap.
Observium has a tendancy to become huge over time with its eventlog containing every port that flaps, syslog, performance timings for every run, and more.
This image comes with a cronjob that will run daily for you to clean this up.
By default it will run with the switches set for eventlog and performance
timings, but like most things in this image it can be overridden by providing
HOUSEKEEPING_ARGS
as an environment variable.
Please checkout out the initial post in the mailing list for details on how to configure your config.php for better control and what each switch does.
Currently the stdout/stderr for the job isn't being sent to /dev/null so you
should be able to see the status of your cleanups with docker logs [id]
It is recommended by Observium to run the housekeeping script manually first before relying on the crons since it may take awhile to first run. Do this if you are using existing data for this image.
This image installs and runs rrdcached. To make use of it, make sure your observium configuration sets
$config['rrdcached'] = "unix:/var/run/rrdcached.sock"