Homelab Monitoring Made Easy - Part 1: Tools Overview - Grafana, Prometheus, InfluxDB, Telegraf

[@linsihan2719]

just wanna come to say thank you! I've been watching your Homelab series to build up my home server, really appreciate your contents here!

[@Jims-Garage]

Thanks, appreciate your feedback 🙂

[@yas-k2j]

Thanks Jim, your videos always has great explanation and easy step by step solutions.

[@Jims-Garage]

Wow, thank you for the kind donation!

[@luisliz]

Been loving your videos. This came out just in time, i'm finally getting to the monitoring part of my homelab. TY SO MUCH!

[@Jims-Garage]

Thanks 👍 I'm hoping to have part 2 out shortly.

[@kevinhughes9801]

Brill stuff great to see your sub count growing defo deserved

[@Jims-Garage]

Thank you for your support

[@octaviusss]

Great video! thank you so much! Subscribed!

[@Jims-Garage]

Appreciated, thanks 👍

[@fedefede843]

Nice! I have a similar setup at home (using a RPi4). Grafana, Loki and Promtail. But instead InfluxDB I use Prometheus. I found it better fit to manage infrastructure (in any case for my home usage it would be the same :)). Along Prometheus I use Node Export for the node metrics and cAdvisor for the containers. I see here with Telegraf you get both in one go. Nice. InfluxDB I use it for Home Assistant historic data. I find it better for that scenario. Long time period retention and lot of sensors going on. At work I am currently trying to deploy a distributed stack, where I use Prometheus in agent mode in the servers, writing out to a central Prometheus instance having all our metrics in one place. On top of that Grafana. Thanks for the video!

[@Jims-Garage]

Thanks, sounds like you have a detailed setup 👍

[@issamzgybi9761]

THANK YOUUUUU JIM

[@Jims-Garage]

@@issamzgybi9761 you're welcome

[@HunterGeophysicsAustralia]

At 21:50, the default username and password is "admin". Hope this helps someone in future...

[@mnejmantowicz]

Jim, I’ve recently discovered your channel. Great content with clearly articulated details and gotchas. Well done sir. I do have a question in regards to this video, if I create a docker stack containing all the services and I need Telegraf to scrape an external resource outside the docker network, won’t it fail in this setup? I always seems to get “no route to host” or the like when I try to connect externally from within the container.

[@Jims-Garage]

Thanks for the feedback. It should be fine to link to external services, I do. Need to make sure that there're no firewall rules in the way and that the correct ports are open on the host

[@mnejmantowicz]

@@Jims-Garage I appreciate the reply. I’ll investigate further. 👊🏻

[@nwdsc]

Great presentation. I'm wondering how resource intensive this stack is. In a limited environment, will this be a CPU or memory hog? Especially once you get numerous monitors configured.

[@Jims-Garage]

Thanks, I recommend that you first cut the stack to only what you need. Then, impose resource limits on the containers, and only record what you need at a sensible frequency and retention period. I'm reasonably confident that it should remain lightweight.

[@funashimi]

hey! thanks for the tutorial! got a couple of problems with your docker-compose in my synology NAS, hope it helps some viewers 1) in Jim's docker-compose ports are labeled differently from container to container, some are with double quotes: - "xxxx:xxxx" others without: - xxxx:xxxx I tried to compose up this in portainer and no ports were published in the end. So I just deleted double quotes so every port is declared in the same way and it worked! 2) during prometheus setup i got 2 errors: - err="open /prometheus/queries.active: permission denied" - err="open /etc/prometheus/prometheus.yml: permission denied" i tried to troubleshoot them myself for hours, but in the end I saw your "user: 0" with grafana container and added that to prometheus too. not elegant and security risk, but chmodding it to 777 is worse i suppose?...

[@Jims-Garage]

Thanks 👍 user 0 is worse as it runs the container as root (I think). 777 simply means anyone can alter it (but they're not root)

[@grocerylist]

How and where would I find the global id of my docker user?

[@Jims-Garage]

Hey, it's id -u username and id -g username

[@grocerylist]

@@Jims-Garage Maybe I missed something. Am I supposed to create a new user for this Monitoring docker stack?

[@dereksloan3233]

@@grocerylist Did you get the docker user global id figured out?

[@yonko1941]

Hi, I do not have reverse proxy running. Which port should I access Grafana? I tried adding a port in the docker compose file but its not working :/

[@Jims-Garage]

@@yonko1941 I think it's 3000

[@yonko1941]

@Jims-Garage neither :(

[@JPEO]

After I've run the docker compose and everything is pulled, I got message in the end [network proxy delared as external, but could not be found] did I missed something? Thanks for the video by the way.

[@Jims-Garage]

This is because I'm using an external network called proxy that I created when I deployed Traefik. Check out my Traefik video if you need further help.

[@JPEO]

@@Jims-Garage Thanks for you prompt reply, unfortunately I cannot use reverse proxy because I cannot do port forwarding.

[@Jims-Garage]

@@JPEO Sure you can, follow my Traefik guide. You use a DNS challenge method, not a TLS. This doesn't require port forwarding.

[@JPEO]

@@Jims-Garage I didnt know that I'll watch it again.

[@JoshArchers]

Great video - but InfluxDB has really put pay to wanting to mess with the TICK stack, or any other variation of it, what with the 3.0 Edge and Community shenanigans. I think these days it would be far better for the homelab market to use Netdata, or perhaps just Prometheus and Grafana together.

[@Jay.M]

The configuration does not work and there is no instruction on how to get it running, you are missing a few configuration files (e.g. loki-config.yml) so those services wont start. Information on that would be helpful. Also, the Grafana container doesn't have the port assigned to it so that starts but is inaccessible.

[@Jims-Garage]

Perhaps things have changed but it worked at the time of recording, I'll check. I'm using a reverse proxy so no ports are required for grafana, it's much nicer using fqdn with Https. Add ports if you don't want to use a proxy.

[@gabrielpetry4099]

i didn't get the idea on why using Prometheus with you are already using influx that has all the things, promtail there is the tail thing for influx for example

[@andrebrait]

Influx doesn't do alerting. Hence why Prometheus is used.

[@raul230285]

Nice