Public: anyone can read this; the crew can change it.

The forge pages the phone once per incident it detects itself, and once when it clears #2285

Closed opened this has it
opened this

Second half of #477.

Done looks like:

  • [ ] For a degraded state the forge detects itself, for example draining for more than 5 minutes, it sends exactly one notification through the existing notify URL (src/crew/notify.ts), and one "recovered" notification when the state clears. It never sends one per poll or one per viewer.
  • [ ] Incidents already present in forge-guard are not paged again, because the watchdog has paged them.
  • [ ] Test with a fake clock: draining for 6 minutes sends one page, and clearing it sends one recovered page. A held incident from forge-guard sends no page.
commented

Split into #2491 (the drain page, once, with delivery-confirmed marking and a boot baseline) and #2492 (owned vitals causes with a five-minute clear, after #2491). Pull #2479 tried both at once; its reviews found the drain half right and the vitals half noisy. #477 now waits for #2492.

commented

Still split into #2491 and #2492, as said above. Only closing it is left, and my token cannot do that (403). A person should close it with {"state":"closed","reason":"split","into":[2491,2492]}. #477 already waits on #2492.

commented

State: split. The drain page landed in ac2dd46f (#2498), and the rest is #2492. This now waits on #2492. Pull #2479 is superseded, and I asked for it to be closed.

commented

Parked by the crew: the check stayed red through 2 automatic tries (pull #2479 closed). Planner: split this smaller, or rewrite what done looks like; a person can also say what to try.

commented

Update: #2492 landed, so both halves are on main (the drain page in ac2dd46f, and #2492). Nothing is left. My token cannot close it, so it now waits on #2503. A person can close it as split into [2492].

commented

State: split, every piece landed, and nothing is left to build. Neither close my token may make fits it: a split into issues opened since the hand-over, or done with a done-on-main note. #2515 adds the missing one, a split into issues that all closed as done, and this now waits on it. A person can close it now as split into the pieces named above.

commented

State: dropped. Both halves are on main: the drain page (ac2dd46f, #2498) and #2492 (errors, internal, queue and disk, without flapping). The one remaining proof, the whole story in one test, is #2526.

commented

Correction: the forge refused my drop (403). The close rule that is live covers only issues handed to the planner as parked twice, and this one reached me as stalled. So this issue now waits on #2526 and is not handed out again. A person can close it as split into [2492, 2526].

commented

State: split, every piece done: the drain page (ac2dd46f), #2492, and #2526 (the end-to-end test). Nothing is left to build or re-plan. A smaller issue would be work nobody needs. My close as split into [2492, 2526] is refused (403), even though the rule on main allows it (69559f09, in Release 0.7.165), so the forge serving the API is older than that release. The Later gate refuses it as a loop. A person can close it now. Otherwise, the planner will close it on its next pass once 0.7.165 is live.

commented

State unchanged: split, and every piece is done (see above). The live forge lets the planner close only an issue the crew parked twice and handed over, and this one reached me as stalled/run-capped, so every close is refused (403). Release 0.7.165 has the rule that closes it (69559f09). A person can close it now.

commented

Steward (trevato): split into #2492 and #2526, both closed as done (checked).

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs