A recovery never waits behind background work: a ready standby first, and an urgent read interrupts a background one #77

Merged
cmoriarty merged 2 commits from recovery-first into develop 2026-11-01 11:53:25 -05:00
Owner

What happened (2026-11-01, Racing vs. Real Madrid, watched in Emby)

  • 16:27 UTC: topsurge.st's stream stopped sending segments.
  • The recovery's first step, a fresh read of the same link, waited for the read gate's one slot. That slot was held by a background check of Villanova vs. Notre Dame, whose links are all dead, walking five links. It ended at 16:29:13.
  • The re-read took 20 s and landed back on topsurge at 16:29:33: 2 min of filler.
  • A verified standby was ready the whole time (givemestreams.cfd, 1920×1080, 59.94 fps, a different stream). Taking it needs no site read.

What changes (recovery-first)

  • A ready standby goes first in a recovery, before any re-read. The failed link cools. With no standby ready, the link is re-read as before. A relapse within 5 min (#65) is still named in the event.
  • An urgent read never waits behind a background read. Finding every slot held, it interrupts the background read that started last and takes its slot.
    • The interrupted check is logged as "check stopped for a watched game's recovery". It isn't the game's failure, and a later round resumes it.
    • An interrupted renewal (#73) keeps its one try.
    • An urgent read still waits for another urgent read.

Checks

  • make test: 853 passed.
  • openspec validate recovery-first --strict passes.

To merge with #76 and deploy once nobody is watching.

🤖 Generated with Claude Code

## What happened (2026-11-01, Racing vs. Real Madrid, watched in Emby) - **16:27 UTC:** topsurge.st's stream stopped sending segments. - **The recovery's first step**, a fresh read of the same link, waited for the read gate's one slot. That slot was held by a background check of Villanova vs. Notre Dame, whose links are all dead, walking five links. It ended at 16:29:13. - **The re-read took 20 s** and landed back on topsurge at 16:29:33: **2 min of filler**. - **A verified standby was ready the whole time** (givemestreams.cfd, 1920×1080, 59.94 fps, a different stream). Taking it needs no site read. ## What changes (`recovery-first`) - **A ready standby goes first in a recovery,** before any re-read. The failed link cools. With no standby ready, the link is re-read as before. A relapse within 5 min (#65) is still named in the event. - **An urgent read never waits behind a background read.** Finding every slot held, it interrupts the background read that started last and takes its slot. - The interrupted check is logged as "check stopped for a watched game's recovery". It isn't the game's failure, and a later round resumes it. - An interrupted renewal (#73) keeps its one try. - An urgent read still waits for another urgent read. ## Checks - `make test`: 853 passed. - `openspec validate recovery-first --strict` passes. To merge with #76 and deploy once nobody is watching. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
A recovery never waits behind background work: a ready standby first, and an urgent read interrupts a background one
All checks were successful
ci / Unit tests, type check, bundle (pull_request) Successful in 50s
ci / One container, e2e click-to-play inside the image (pull_request) Successful in 1m25s
e22ee5fb27
On 2026-11-01 Racing vs. Real Madrid sat 2 min on filler: topsurge froze,
and the recovery's re-read waited for the one read slot, held by another
game's five-link background check, while a verified 1080p standby was
ready.

- A recovery takes a ready standby first (no site read); with none
  ready it re-reads the link as before. A relapse (#65) is still named.
- An urgent read that finds every slot held interrupts the background
  read that started last. The interrupted check is logged, isn't the
  game's failure, and a later round resumes it; an interrupted renewal
  keeps its try.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Merge remote-tracking branch 'origin/develop' into recovery-first
All checks were successful
ci / Unit tests, type check, bundle (pull_request) Successful in 43s
ci / One container, e2e click-to-play inside the image (pull_request) Successful in 1m21s
33c3bd9232
cmoriarty merged commit a3b195a4b3 into develop 2026-11-01 11:53:25 -05:00
Sign in to join this conversation.
No reviewers
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
cmoriarty/sports-breaker!77
No description provided.