Metadata Valid
No translation
Allow specifying other services that should be started prior to the exporter daemon starting up in systemd. This is to handle situations where you want the service being monitored to be started before the exporter.
N/A (new feature)
<!--
Thank you for contributing to this project!
-->
Add package prometheus-node-exporter package name for Debian.
Installation on Ubuntu 18.04
n/a
…orter
<!--
Thank you for contributing to this project!
-->
<!--
Replace this comment with a description of your pull request.
-->
<!--
Replace this comment with the list of issues or n/a.
Use format:
Fixes #123
Fixes #124
-->
Fixes #448
Fixes #434
Note: this PR is still Work-In-Progress as I have not tested all edge cases but it does work. Perhaps you have the time to take a look on your side and maybe your will either immediately tell me that is totally wrong or the other way around. I will get back to this when I will have more time.
This PR is intended to allow environment variables to be set for the blackbox exporter.
This can be useful, for example, if you want to use SSL_CERT_FILE.
Fixes #535
The "name" block is an object. This actually matches the example in
prometheus::alertmanager's documentation:
```
```
I wonder why unit tests didn't catch those...?
Closes: #539
<!--
Thank you for contributing to this project!
-->
<!--
Replace this comment with a description of your pull request.
-->
<!--
Replace this comment with the list of issues or n/a.
Use format:
Fixes #123
Fixes #124
-->
With larger WAL segments, prometheus fails to write
a new checkpoint in TimeoutStopSec time:
Oct 03 15:13:37 prometheus prometheus[2452]: level=info ts=2020-10-03T15:13:37.751Z caller=checkpoint.go:96 component=tsdb msg="Creating checkpoint" from_segment=85417 to_segment=85677 mint
Oct 03 15:15:06 prometheus systemd[1]: prometheus.service: State 'stop-sigterm' timed out. Killing.
Oct 03 15:15:06 prometheus systemd[1]: prometheus.service: Killing process 2452 (prometheus) with signal SIGKILL.
Oct 03 15:15:06 prometheus systemd[1]: prometheus.service: Main process exited, code=killed, status=9/KILL
Oct 03 15:15:06 prometheus systemd[1]: prometheus.service: Failed with result 'timeout'.
Oct 03 15:15:06 prometheus systemd[1]: Stopped Prometheus Monitoring framework.
This change adds a new parameter to define TimeoutStopSec
for prometheus.service.
No issue previously created.
The required systemd unitfile overwrite could have been made locally as well - but as I suppose that others could also face similar issues, I decided to implement the parameter directly.
This is the result of using sed + rename. Untested, may eat your dog etc.
Together with #430 this PR adapts the Debian-defaults regarding package- and service-names like they are actually present in Debian. e.g. all exporters are prefixed with prometheus-.
https://packages.debian.org/search?keywords=prometheus&searchon=names&suite=stable§ion=all
```bash
user@host:~$ for A in $(apt-cache search prometheus- | awk '{ print $1 }' | xargs); do apt-file show $A 2>&1 | grep -E '(systemd/system)'; done
prometheus-alertmanager: /lib/systemd/system/prometheus-alertmanager.service
prometheus-apache-exporter: /lib/systemd/system/prometheus-apache-exporter.service
prometheus-bind-exporter: /lib/systemd/system/prometheus-bind-exporter.service
prometheus-bird-exporter: /lib/systemd/system/prometheus-bird-exporter.service
prometheus-blackbox-exporter: /lib/systemd/system/prometheus-blackbox-exporter.service
prometheus-haproxy-exporter: /lib/systemd/system/prometheus-haproxy-exporter.service
prometheus-mongodb-exporter: /lib/systemd/system/prometheus-mongodb-exporter.service
prometheus-mysqld-exporter: /lib/systemd/system/prometheus-mysqld-exporter.service
prometheus-nginx-exporter: /lib/systemd/system/prometheus-nginx-exporter.service
prometheus-node-exporter: /lib/systemd/system/prometheus-node-exporter-apt.service
prometheus-node-exporter: /lib/systemd/system/prometheus-node-exporter-apt.timer
prometheus-node-exporter: /lib/systemd/system/prometheus-node-exporter-ipmitool-sensor.service
prometheus-node-exporter: /lib/systemd/system/prometheus-node-exporter-ipmitool-sensor.timer
prometheus-node-exporter: /lib/systemd/system/prometheus-node-exporter-mellanox-hca-temp.service
prometheus-node-exporter: /lib/systemd/system/prometheus-node-exporter-mellanox-hca-temp.timer
prometheus-node-exporter: /lib/systemd/system/prometheus-node-exporter-smartmon.service
prometheus-node-exporter: /lib/systemd/system/prometheus-node-exporter-smartmon.timer
prometheus-node-exporter: /lib/systemd/system/prometheus-node-exporter.service
prometheus-pgbouncer-exporter: /lib/systemd/system/prometheus-pgbouncer-exporter.service
prometheus-postgres-exporter: /lib/systemd/system/prometheus-postgres-exporter.service
prometheus-process-exporter: /lib/systemd/system/prometheus-process-exporter.service
prometheus-pushgateway: /lib/systemd/system/prometheus-pushgateway.service
prometheus-snmp-exporter: /lib/systemd/system/prometheus-snmp-exporter.service
prometheus-sql-exporter: /lib/systemd/system/prometheus-sql-exporter.service
prometheus-squid-exporter: /lib/systemd/system/prometheus-squid-exporter.service
prometheus-varnish-exporter: /lib/systemd/system/prometheus-varnish-exporter.service
prometheus-xmpp-alerts: /lib/systemd/system/prometheus-xmpp-alerts.service
```
<!--
Thank you for contributing to this project!
-->
<!--
Replace this comment with a description of your pull request.
-->
Allow the default scrape configs for Prometheus to be disabled.
In my environment I will make use of this in Hiera
lookup_options:
prometheus::scrape_configs:
merge: unique
With the above and current module behavior, I would be forced to use this module's default scrape config for Prometheus which I don't want because we define the Prometheus scrape config differently. This pull request allows for not including that default scrape config so that something like Hiera merging isn't forced to use this module's scrape config.
<!--
Thank you for contributing to this project!
-->
It's desirable to be able to specify Puppet fileserver urls (puppet://...) as the download_url for exporters in cases where exporter releases are mirrored internally to avoid reliance on reaching the internet.
n/a
<!--
Thank you for contributing to this project!
-->
Support mutetimeintervals for Alertmanager config
If user defines something like:
prometheus::alertmanager::mute_time_intervals:
- name: business_hours
time_intervals:
- weekdays: [ 'monday','tuesday','friday' ]
times:
- start_time: '16:00'
end_time: '23:59'
- start_time: '00:00'
end_time: '07:00'
it won't be picked up, and therefore create a configuration error on other places "missing definition for business_hours".