Hi all. I just set-up my first self-hosting server with NextCloud, Immich and a VPN server. I was wondering if there is a tool or layer of tools which would help me monitor my server and the services including running stats, resource usage stats, system logs, access logs, etc?

I read that Grafana Loki along with Prometheus could possibly help me with this. I just wanted to ask that - should I explore these two tools or do we have some other and better(suiting to my needs) tools? Please recommend Open Source tools only. Preferably Docker, or Linux based otherwise. Thank you :))

all 18 comments

sorted by: hot top controversial new old
[–] 16 points 2 years ago (2 children)
  • [–] 4 points 2 years ago (1 child)
  • [–] 9 points 2 years ago (2 children)

    -1 for Netdata. I used it for a bit, but the configuration is not very intuitive and the docs for alerts were basically “rest of the fucking owl”, at least for the non-cloud version. I ended up just switching to Glances which is pretty boneless but it’s easy.

    Though for OP I’d probably recommend Prometheus.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 7 points 2 years ago (1 child)

    Also -1 for netdata. I loved the analytics but it brought all of my VMs to a screeching halt. It did not seem very will optimized for the amount of data it was polling.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 2 points 2 years ago

    I went through setting up netdata for a sraging (in progression for a production) server not too long ago.

    The netdata docs were quite clear on that fact that the default configuration is a "showcase configuration", not a "production ready configuration"!

    It's really meant to show off all features to new users, who then can pick what they actually want. Great thing about disabling unimportant things is that one gets a lot more "history" for the same amount of storage need, cause there are simply less data points to track. Similar with adjusting the rate which it takes data points. For instance, going down from default 1s internal to 2s basically halfs the CPU requirement, even more so if one also disables the machine learning stuff.

    The one thing I have to admit though is that "optimizing netdata configs" really isn't that quickly done. There's just a lot of stuff it provides, lots of docs reading to be done until one roughly gets a feel for configuring it (i.e. knowing what all could be disabled and how much of a difference it actually makes). Of course, there's always a potential need for optimizations later on when one sees the actual server load in prod.

  • source
  • parent
  • [–] 3 points 2 years ago*

    This isn't specific to just netdata, but I frequently find projects that have some feature provided via their cloud offering and then say "but you can also do it locally" and gesture vaguely at some half-written docs that don't really help.

    It makes sense for them, since one of those is how they make money and the other is how they loose cloud customers, but it's still annoying.

    Shoutout to healthcheck.io who seem to provide both nice cloud offerings and a fully-fledged server with good documentation.

  • source
  • parent
  • I'm a big fan of netdata; it's part of my standard deployment. I put in some custom configs depending on what services are running on what servers. If there's an issue it sends me an email and posts into a slack channel.

    Next step is an influxdb backend to keep more history.

    I also use monit to restart certain services in certain situations.

  • source
  • parent
  • [–] 12 points 2 years ago

    I like Uptime Kuma, but it only monitors if a service is online or not. I'm up to 21 services now so I'm not interested in all their details, just if I need to fix something urgently.

  • source
  • [–] 7 points 2 years ago

    Grafana + Prometheus + data gathering will at least give you the resource and usage stats.

  • source
  • [–] 5 points 2 years ago (1 child)

    I’m using Zabbix to monitor the most important bits.

  • source
  • hideshow 2 child comments
  • [–] 4 points 2 years ago

    I'd swap Prometheus for VoctoriaMetrics. It's a drop-in replacement with a much better resource consumption story and a few extra goodies.

  • source
  • [–] 3 points 2 years ago

    Grafana + Prometheus dashboards can be quite addicting or useful. Noted.lol put together a nice tutorial for getting started.

    For most of my services though, I simply use Uptime Kuma which then sends an alert to Gotify when my services go down or whatnot, Gotify then instantly notifies my phone so I can be aware. It helps keep the spouse happy when their go to service for some reason crashed. :)

  • source
  • [–] 2 points 2 years ago (1 child)

    How much do you point and click your setup?

    As I use nagios

  • source
  • hideshow 2 child comments
  • [–] [S] 1 point 2 years ago (1 child)

    I'm a web-app developer myself. So I don't mind configuring things if needed. I can opt to configure if it meets my goals better. I'd check out nagios. :))

  • source
  • parent
  • hideshow 2 child comments
  • [–] 2 points 2 years ago* (last edited 2 years ago) (1 child)

    I've not found a good solution for actual constant monitoring and I'll be following this thread, but I have a similar/related item: I use healthcheck.io (specifically a self-hosted instance) to verify all my cron jobs (backups, syncs, ...) are working correctly. Often even more involved monitoring solutions do not cover that area (and it can be quite terrible if it goes wrong), so I think it'll be a good addition to most of these.

  • source
  • hideshow 2 child comments