Index of /Monitoring
Name Last modified Size
Parent Directory -
read_d3p_temperature..> 2012-12-03 21:21 1.7K
ReadMe.txt 2014-06-22 23:18 4.1K
sqlite3stats.pl.txt 2012-10-01 22:35 5.9K
read_d3p_temperature..> 2012-12-03 22:36 7.4K
cambridge-dtg.pl.txt 2011-09-19 18:41 11K
plc_.102.txt 2013-08-25 00:47 12K
plc_.103.txt 2014-02-01 20:42 17K
diskstatjf_ 2014-02-09 13:59 20K
diskstatjf_.txt 2014-02-09 13:59 20K
smartstats_.txt 2014-06-22 23:18 22K
plc_.txt 2014-02-01 20:42 23K
Monitoring
==========
I have various things monitored at home about the server I run. As I was
very concerned about the temperature of the server during the summer a
year or so back I began monitoring temperatures as well - I've used Day
Barr's plugins for monitoring now that the Dell desktop I use as my
server doesn't have any internal monitoring.
Anyhow, if there are any interesting plugins that I create, I'll put
them here. Eventually there should be a few that I've written which get
moved here in case they're useful to anyone else.
cambridge-dtg:
The Cambridge Digital Technology Group have some weather sensors
on the top of their labs. As they're accurate for Cambridge, where
I live at present, I use them as monitors for the general environment.
They give a reasonable indication of the temperatures at the time.
This can then be used alongside the server-specific temperature to
get some useful information.
The plugin can function either in text mode (where it really just
re-prints the textual information from their site), as an MRTG plugin,
or as a munin plugin.
sqlite3stats.pl
It is often useful to track the usage of databases. MySQL and
PostgreSQL have plugins to monitor many parts of their usage, but
I could not find any similar plugin for SQLite databases. I wrote a
simple plugin that will produce some statistics about the database.
At present we just report the size of the file, and the highest
row id for each table, which approximates the number of rows.
read_d3p_temperature.sh
A friend of a colleague built a custom temperature sensing device.
It's a USB interface, with 8 temperature sensors, on the end of
2 meters of cable each. It's very simple to communicate with, so I
wrote a simple shell script to do so.
read_d3p_temperature.pl
Having written the simple script to just read the values, I wrote a
more complex script that allows the reading of temperatures as human
readable format, for MRTG or for Munin, and can be configured to run
with different parameters in each case. Probably overkill, but it
appears to work.
plc_
A munin plugin which can monitor the Powerline network speed and
signal to noise ratio. I use this with my TP-Link powerline network
devices. Revised (v1.04) to understand both AR6400 and INT7400
devices. Thanks to Alex Macfarlane-Smith for testing.
plc_102
Older Munin plugin, which did not correctly understand the way in
which the network topology was described, but which is a little
simpler (and should work correctly if you don't reset the devices on
the network).
plc_103
Older Munin plugin, which only handled more recent Atheros devices
(those which used the 'amp*' tools).
diskstatjf_
A modification of the standard diskstat_ data collection plugin
to key the drives by their model and serial number, instead of by
the device name. In a system that has multiple controllers, and
swappable discs, the device names - which are dynamically assigned
for the 'sd' devices - can end up in a different order to that
which they were originally. This means that the historic data
for each device is not related to the device name - and as discs
are moved around, or the controllers appear in a different order
on startup, the historic records for each device would be
incorrectly assigned. Using the model and serial number should
keep the historic records accurate. I also added a title to each
drive which gives the labels that were assigned to the filesystems
on the drive. Some drives may not have labels (and in any case
they are assigned to the FS, not the drive) so the label is not
used as a distnguisher.
smartstats_
A munin plugin designed to report on the 'sataphy' statistics reported
by the logs in modern discs. Requires 'smartctl'. Uses the same logic
to select drives as the the above 'diskstatjf_' tool. I use the tool
configured with 'ACTIVEONLY' set to '1', and 'CACHE' of '120', so that
we only wake up discs that would be awake anyhow.
Updated 2014-05-13 to add support for reading SSD usage information.
Updated 2014-06-22 to use udevadm in preference to udevinfo.
--
Justin Fletcher