xperimental/nextcloud-exporter

Multi instance monitoring

smycek opened this issue · 8 comments

smycek commented

Do you have some recommendations or best practice when setting up monitoring multiple Nextcloud Instances?

xperimental commented

I personally have no experience with monitoring a large Nextcloud installation, because I only host relatively small ones.

Exporters usually have a 1:1 relationship with the service they are monitoring. In this case this could be interpreted as one of the following:

  • one exporter instancer per server hosting the Nextcloud instance
  • one exporter per "logical Nextcloud instance"

I'm leaning towards the second one, because the exporter mostly monitors metrics of the Nextcloud installation and not the underlying server, for which there should be other exporters available already.

  • 👍1
  • ❤️1
rtt-mmk commented

We run a nextcloud installation of 10 Nextcloud-Apache-Hosts in front of DB- and Redis-Clusters.

For the monitoring we have an additional host fully equipped with nextclouds code. On this node we query the serverinfo.php to write the output into our time-series-database.

Detailed metrics (OS, Filesystems etc.) of the servers we collect with an extra server-agent on all servers, that indepently writes this metrics into our monitoring-database.

I can confirm, that querying the nextcloud endpoint only from one server is suitable in larger setups, too. Although at some scaling point this will ran into some issues and timeouts with the storage-countings they implemented there. But it reports the metrics for the whole "logical Nextclound instance".

Because of the scaling issues, we experimenting now with this nextcloud-exporter, instead of running the serverinf.php-polling locally. :)

hoamer commented

Hi,

I just want to share my setup, with "multi" (3 instances).
For example, config files are thrown in "/etc/prometheus_nextcloud_exporter/".

Then I created this unitfile, which is suitable for multiple instance of the nc-exporter: /etc/systemd/system/prometheus_nextcloud_exporter@.service

[Unit]
Description=%i service for prometheus exporter for nextcloud
Requires=nginx.service
After=nginx.service

[Service]
Type=simple
RemainAfterExit=true
ExecStart=prometheus_nextcloud_exporter -c /etc/prometheus_nextcloud_exporter/%i.yml

[Install]
WantedBy=multi-user.target

where

systemctl start prometheus_nextcloud_exporter@nextcloud_instance_1

i.e. is

/etc/prometheus_nextcloud_exporter/nextcloud_instance_1.yml
xperimental commented

@hoamer The service unit in the contrib/ folder sets some additional parameters to improve security, but it can also be easily updated for instancing like you did.

@smycek Has any of the existing posts answered your questions? Would it be okay to close the issue?

jhjacobs81 commented

this is actually just what i ws looking for! awesome!

  • 👍1
rtt-mmk commented

Just for info: To cope with the scaling issues while counting files-statistics in large installation nextcloud tranferred theses queries to a long-running background cronjob (since Version 21.0.2 - nextcloud/serverinfo#298).

smycek commented

@smycek Has any of the existing posts answered your questions? Would it be okay to close the issue?

My question has been answered, I am fine if you close the issue. Thanks to everybody!

xperimental commented

Great. Thanks for the input, everyone.