1. Last account update age in hours
2. Account balances
The plugins have been in use for several months now and warn
when the balance of accounts drop below a configurable value,
and when Hibiscus stopped updating data because a TAN needs to
be entered manually after 90 days.
The plugins directly query the MySQL database of a Hibiscus server
instance.
See https://www.willuhn.de/products/hibiscus-server/ for more information.
The utilization used to be on a line beginning with "Gpu", but has since
been capitalized to "GPU". There's already several other instances of
that in the nvidia-smi output, so narrow down the search by returning
only the line after "Utilization".
The graphs are basically identical to the original Ruby plugin, but
there is no longer a dependency on the "gpspipe" utility; a direct TCP
connection to gpsd is established, and the JSON is parsed using the
json-c library.
The intent here is to provide the same functionality on systems that are
resource-constrained and/or sensitive to jitter (e.g., an embedded NTP
server).
This can be compiled like such after editing GPSD_HOST / GPSD_PORT:
gcc -o gpsd gpsd.c -ljson-c
* [jellyfin_sessions] A simple plugin to check active sessions on a jellyfin media system
* Update plugins/jellyfin/jellyfin_sessions
Co-authored-by: Kenyon Ralph <kenyon@kenyonralph.com>
---------
Co-authored-by: Sebastian L. <sl@momou.ch>
Co-authored-by: Kenyon Ralph <kenyon@kenyonralph.com>
* feat: add cpu_by_group plugin
The current cpu_by_process plugin says it self it can generate huge graphs which are difficult to decipher. This plugin tries to mitigate that issue by automatically grouping processes by their cgroup. This also solves the issue that cpu_by_process groups processes by process name, which means atm you cant really differentiate between f.e. python processes in various cgroups. The remaining processes that are not running in a cgroup are also split between kernel threads and system processes
For user & docker cgroups a human readable cgroup label is constructed (if munin has f.e. docker access). The function in this plugin could probably be expanded for other often used cgroups
* chore: add example graphs & update docs
* chore: update docs
* syslog_ng_stats: mark total counter stats as DERIVE / min 0
the data produced by the underlying `syslog-ng-ctl stats` gives the stats values in absolute numbers since the start of the syslog-ng daemon. In order to visualize in a more user friendly way those values are now marked in the config as type DERIVE.
* syslog_ng_stats: graph_vlabel set as messages/graph_period
This argument print unit dependencies as a list instead of a tree and
avoid showing sometimes colored dot. This dots are displayed to identify
failed units but messed with the awk command, so this failed unit is
never seen. With this --plain argument, dots are not displayed anymore.
Change value from "current absolute value" to "15min average value" and
tune warning limits accordingly. Healing value can sometimes be 1 and
gluster will fix it by itself. Send warning only if it stays at 1 (or
above) for 15 minutes. This will still send a warning immediately if it
goes to 3 or above.
