No metrics for what the daemon holds between requests, such as database connections in use #194
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Logs and traces show requests, scans and downloads, but nothing shows the daemon's own state over time: database connections in use against the pool's size, commands waiting for the worker, live-update listeners, torrent slots, the runtime's tasks, memory and open files. A pool run dry, as the 5 s POST /api/fetch of 2026-10-06 looked like, can only be guessed at. Add a Prometheus /metrics, scrape it, and graph it on the iPX dashboard.
Done and deployed: c2... see git log. /metrics is served by [web] metrics_bind on a thread, runtime and port of its own (192.168.1.130:9187, LAN only, no sign-in), from memory alone; Prometheus scrapes it as job ipx in 1.9 ms (149 ms when it shared the web server and queried the database). The iPX dashboard has a Daemon row; the prod-check skill a prom command and a sixth check. The API token made for the first version is revoked.
Commits, for the record:
9fc1e7e(first version on the web port),973eeb6(own thread, runtime and port, from memory),a985cfa(prod-check skill).