Files
pulse/internal
rcourtman 0de21d8931 Guard Docker update alerts against the daily resolve loop
On v6.3.2 an image update that stayed asserted resolved and re-opened
about every 24 hours (#2353). The hourly sweep expired update tracking
by first detection, the next report restarted the delay below its
threshold without refreshing the open alert, and the stale-alert sweep
resolved it a day later, just before the restarted delay re-fired it.

Two later changes each close a path to that loop: tracking expiry now
follows the latest observation (f7a43d2794), and an open alert's start
stands in for tracking a restart lost (ed8415ea6a). Existing tests check
the tracking maps; this one drives several simulated days of reports and
hourly housekeeping and asserts at the alert itself: one opening, no
resolution, a stable start, while a current image or a removed container
still resolves. Reverting either change fails its own case.
2026-10-16 23:18:51 +01:00
..
2026-10-11 22:20:25 +01:00
2026-10-11 22:20:25 +01:00