Hey everyone, I'm curious about how you're managing Docker container monitoring and alerting in production without racking up massive costs for observability tools. I've noticed a few issues common among small teams and self-hosted setups:
- The command 'docker stats' provides some basic data but lacks historical insight.
- While logs can help troubleshoot after an issue arises, they're not particularly useful for early warnings.
- Many monitoring tools seem either overpriced or too complicated for simple Docker setups.
I've been experimenting with a lightweight approach focusing solely on Docker containers which includes:
- Tracking real-time CPU usage, memory, restarts, and health status.
- Setting up straightforward alert rules (like memory spikes or unhealthy container alerts).
- Notifications sent to platforms like Slack, Discord, email, or WhatsApp.
- Quick configuration (just an agent and a dashboard) instead of a complete observability solution.
I'm interested in hearing your thoughts:
1. What metrics do you consider essential for your Docker workloads?
2. Do you prioritize alerts based on container-level or application-level issues?
3. What are the cost-effective yet reliable setups you've found?
If you're interested, I can share the monitoring rule templates we've been working on, including things like out-of-memory warnings and unhealthy loop detection.
4 Answers
If you're looking for a budget-friendly solution, I recommend Uptime Kuma with Pushover for alerts, WUD for container updates, Dozzle for logging, and Oxker for terminal health monitoring. It handles about 30 containers in my setup.
Honestly, I see a lot of people trying to push the idea that paying for basic uptime alerts is normal. It seems odd when monitoring is already tackled pretty well with existing solutions.
I get the whole 'reinventing the wheel' critique. For me, many tools I've tried were too costly or just too bulky for my needs, especially when I only had a few containers. So I created a simpler version that fits my context better.
I tried LGTM but found it a bit too complex. Then I switched to OpenObserve, which has been great for my needs. It offers a lot though the UI could be better. It's still in development, but it's open-source and hits all the right notes for monitoring.
cAdvisor is a solid choice to integrate with whatever monitoring system you prefer if you're exploring options.
I'm checking out cAdvisor, but have been using Kernus for container alerts and monitoring. Seems to work well so far!
That's a really nice and lightweight combination! I've seen Uptime Kuma and Dozzle work well together, but combining them with Oxker is unique. Have you found any gaps in this setup?”