[smtp-relay] Nobble log ingrestion to allow metrics startup

[djjudas21]

What this PR does / why we need it

At the moment it is not possible for the metrics exporter to read /var/log/maillog from the Postfix container without some additional engineering work on log rotation and possibly even shared volumes - discussion in #146 (comment)

As a workaround, this PR causes the metrics exporter to open a fake Postfix logfile from /dev/null, so it can start up without crashing. This means the metrics exporter will not have access to the Postfix event log (it still has access to the Postfix socket) so it will report a reduced set of metrics. At the moment I'm not sure what it will report.

It should still be possible to measure the queue length and ensure the exporter is working properly, but it might fail to report the number of messages transmitted, etc.

Which issue this PR fixes

• fixes [smtp-relay] monitoring enabled causes CrashLoopBackOff #146

Checklist

• Chart Version bumped
• Title of the PR starts with chart name (e.g. [prometheus-couchdb-exporter])

[djjudas21]

@leleobhz what do you think about this PR? 🙂

[leleobhz]

@djjudas21

I think this is a good start, but some logic for log rotating postfix logs or test if FIFO files are compatible with both postfix and kumina collector may be a next step.