You don't need an observability stack yet

Short summary

Small and indie projects on a single host don't need a full observability stack. The article breaks observability into seven pillars and argues that logs, host metrics, and cron-failure alerts are already handled by built-in Linux tools (docker logs, htop, df, crontab MAILTO). Only errors and uptime genuinely require dedicated tools from day one. The unified-observability vendor pitch saves effort at scale, but for one developer and a few hundred users, it solves a problem you don't have yet.

• •Logs, host metrics, and cron failures on a single host are solvable with built-in Linux commands — no stack needed
• •Observability is 4-7 separate concerns, not one category; most don't fire often enough at indie scale to justify a tool
• •Errors and uptime are the two pillars that genuinely need dedicated tools from day one; the rest can wait until you hit an inflection point

Generated with AI, which can make mistakes.