gvengel / exim_exporter

Exports metrics from the exim mail server for consumption by Prometheus.

Exim Exporter for Prometheus

This prometheus exporter monitors the exim4 mail transport server. Stats are collected by tailing exim's log files, counting messages queued on disk, and observing running exim processes.

Installing

Download and run the latest binary from the releases tab.

./exim_exporter

The exporter will need read access to exim's logs and its spool directory. Therefore, it is recommended to run as the same user as the exim server.

Alternately, a simple Debian package is provided which installs the exporter as a service.

dpkg -i prometheus-exim-exporter*.deb

Usage

By default, the exporter serves on port 9636 at /metrics.

The exporter has two modes:

1. The default mode is to process the log files appended to by exim by tailing them. The default is good for Debian/Ubuntu servers which store the logs in /var/log/exim4. Running in this mode on other distributions may require manually configuring the paths.
2. The second mode utilizes the systemd journal, tailing it for any log lines sent with the syslog identifier exim (configurable). This mode can be enabled using --exim.use-journald.

In both modes the exporter will additionally poll your spool directory to determine the length of the mail queue.

See --help for more details. Command line arguments can also be set via environment variable. e.g --exim.mainlog -> EXIM_MAINLOG.

Building

make

Metrics

See example metrics in tests.

exim_messages_total

This stat is calculated by tailing the exim mainlog and returning a counter with labels for each log message flag. An additional label is added for messages marked as completed.

Prom Label	Exim Flag
arrived	<=
fakereject	(=
delivered	=>
additional	->
cutthrough	>>
suppressed	*>
failed	**
deferred	==
completed	Completed

exim_reject_total and exim_panic_total

These stats are calculated by tailing the rejectlog and paniclog, returning counter for the number of lines in each.

exim_processes

This metric returns the number of running exim process, labeled by process state. The state is detected by parsing the process's command line and looking for known arguments. While this method doesn't provide the same detail as exiwhat, that tool is contraindicated for use in monitoring.

Prom Label	Exim State
daemon	parent pid
delivering	exim -Mc
handling	exim -bd
running	exim -qG
other	other

exim_queue

This metric reports the equivalent of exim -bpc. Note, the value is calculated by independently parsing the queue, not forking to exim.

exim_up

Whether the main exim daemon is running.

exim_log_read_errors

This metrics reports any failures encountered while tailing the logs.

Docker

Docker images are available on docker hub.

To enable full functionality the exporter needs access to exim's log files, spool directory, and process list. For example, if you were running your MTA in a container named exim4, usage might look something like:

docker run 
  -p 9636:9636 \
  -v /var/log/exim4:/var/log/exim4 \
  -v /var/spool/exim4:/var/spool/exim4 \
  --pid container:exim4 \
  --name exim_exporter \
  gvengel/exim_exporter

Also see the provided docker-compose example.