The daemon runs a scan pass every minute, nearly always for nothing #114
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?
The scheduler ticks every 60 s (main.rs, ticker) and each tick runs a full scan pass: subscriptions, http_state for every feed (~180 queries) to find which are due, and before it the reap sweep (missing files, quota and age, the Directory cleanup, the artwork cache trim). Most passes find no feed due and fetch nothing: 362 scan_done in 6 h on 2026-10-02, nearly all with feeds=0, each a trace in Tempo and an event to every open page. Wanted: after a pass, sleep until the next feed is due (commands still wake it at once), and run the sweep on its own, slower clock, e.g. hourly.
Fixed in
ec8fd5d: the daemon sleeps until the earliest feed is due (30 s to 10 minutes) instead of a scan pass every minute; commands wake it, and the first pass after start runs at once. One query for every feed's check state. After the 19:57 deploy every scan but the start-up one checked feeds (6, 8, 15, 34), where before 81% checked none.