You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
file-updater: a failed HTTP request never calls back, so ingest refresh waits time out (3-5 s, UI thread in auto-config) #1347
Part of #1273. B, waits out the full timeout. Left over from #1284 (PR #1326).
When the HTTP request for an ingest list fails (network down, DNS, TLS, HTTP error), shared/file-updater never calls the update callback. The rtmp-services refresh-done event (ingest-refresh-wait.h) is therefore never signalled, and the caller waits out the full timeout:
3 s from rtmp_common_url (the output's connect thread);
5 s on the UI thread in the auto-config wizard (AutoConfig::AutoConfig).
A download that arrives but doesn't parse already ends the wait (#1326); only the "no response at all" path is left.
Fix: have file-updater report the end of every attempt to the callback (success, HTTP failure or a network error), or give it a completion hook, so ingest_update / dacast_ingest_update can signal the event on failure too. Keep the existing callback semantics for other file-updater users, such as the services list.
Test: with an unreachable URL (a closed localhost port), the ingest refresh wait ends within about one connect failure, far below the 3–5 s timeout.
Claiming. Plan: file-updater reports the end of every attempt (failure too) to its users through a new completion path, without changing what existing callbacks see on success; rtmp-services signals the refresh-done event from it. RED: a closed localhost port ends the wait within about one connect failure, not 3–5 s.
Activity
zackees commentedon Oct 17, 2026
Claiming. Plan: file-updater reports the end of every attempt (failure too) to its users through a new completion path, without changing what existing callbacks see on success; rtmp-services signals the refresh-done event from it. RED: a closed localhost port ends the wait within about one connect failure, not 3–5 s.
docs: link PR #1353 in the #1347 ledger line
file-updater: Report a failed download so ingest refresh waits end (o…
zackees commentedon Oct 23, 2026
Fixed by #1353 (merged). That PR used
Refs, so the issue stayed open, but its body lists no remaining work.