Ask HN: Light self-hosted logs/metrics/traces dashboard?

[z33k]

I’m aware of enterprise scale observability tools like ELK stack and datadog. I’d like to monitor my containerized (hobby) web app running on a $6/mo VPS. I’d rather self-host to keep the costs low, but grafana takes up gigabytes of memory that my VPS does not have. Are there any lighter alternatives? My hobby site can handle around 500 concurrent users and I don’t expect much growth.

[KronisLV]

For monitoring servers: Zabbix would work.

For APM across some popular stacks: Apache Skywalking might be worth a look (though probably with PostgreSQL instead of ElasticSearch).

For analytics: Matomo is good.

For uptime monitoring: Uptime Kuma is really good.

Consider a separate cheap VPS for the monitoring stack, whatever it might end up being, like something from Hetzner.

[GauntletWizard]

Grafana doesn't take up gigabytes unless you're doing something crazy; it's about 100mb in my small business deployments. Which still might be too much for your $6 VPS! I would still highly recommend Prometheus, as it's native dashboard has gotten a lot better with 3.0

[brokegrammer]

One way to save costs would be by using journalctl to view logs then setup your app to send you an email whenever an unhandled exception is raised or some other kind of error occurs.

[not_your_vase]

Poor man's dashboard: Get your desired metrics with grep/wc/whatever shell commands from your logs, and visualize with PyQtGraph/PySide (or even gnuplot, for an even lighter experience).

[skwee357]

Or you can use GoAccess [0]

[0] https://goaccess.io/

[kingkongjaffa]

interesting

> GoAccess is written in C

I assumed it would be written in Golang from the name.

[vdvsvwvwvwvwv]

I would love to know the answer? I wonder if Prometheus and Grafana would work at this light level of compute?

[PeterZaitsev]

Check out Coroot, it is rather light, simple to deploy and understand

[rickette]

netdata https://github.com/netdata/netdata