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

When checks wait behind one lane, the queue:check card names the setting that would widen it #5131

Parked opened this
Closed without merging

When one lane is the reason checks are waiting, the queue:check card now ends with Checks at once is 1 — raise it in Settings → Capacity. The setting's name comes from CAPACITY["checks.concurrency"].label in laneSentence (src/kernel/vitals.ts), not a copied string.

Notes

stopped parked by the crew deploy da6003b

opened this

When one lane is the reason checks are waiting, the queue:check card now ends with Checks at once is 1 — raise it in Settings → Capacity. The setting's name comes from CAPACITY["checks.concurrency"].label in laneSentence (src/kernel/vitals.ts), not a copied string.

  • Unchanged cases: with more than one slot the sentence is the same as before. The card's href is still /settings/capacity.
  • Why the last check failed: I couldn't reproduce it. "the sentinel command > an admin has it, and it says what the board says" (test/tui-sentinel.test.ts) passed three runs in a row here, both before and after merging current main. That test doesn't touch laneSentence, so I haven't tied the failure to this change. I haven't read the failing log.
  • What I did: merged current main into the branch, since force pushes are refused. The vitals, red-main, sentinel and tui-sentinel tests pass after the merge: 78 pass, 0 fail.
  • Tests: the new vitals test checks that one slot names the setting and three slots don't. I updated the old assertions that expected the previous sentence ending, in test/vitals.test.ts and test/red-main.test.ts.
  • Left alone: the sentinel.test.ts fixture row that holds the old sentence is stored input, not an assertion on the new text.
  • If it fails again: the next failure's log should show whether that test is flaky on its own or this change really interacts with it.

Closes #5111

commented

When one lane is the reason checks are waiting, the queue:check card now ends with Checks at once is 1 — raise it in Settings → Capacity. The setting's name comes from CAPACITY["checks.concurrency"].label in laneSentence (src/kernel/vitals.ts), not a copied string.

  • Unchanged cases: with more than one slot the sentence is the same as before. The card's href is still /settings/capacity.
  • Why the last check failed: I couldn't reproduce it. "the sentinel command > an admin has it, and it says what the board says" (test/tui-sentinel.test.ts) passed three runs in a row here, both before and after merging current main. That test doesn't touch laneSentence, so I haven't tied the failure to this change. I haven't read the failing log.
  • What I did: merged current main into the branch, since force pushes are refused. The vitals, red-main, sentinel and tui-sentinel tests pass after the merge: 78 pass, 0 fail.
  • Tests: the new vitals test checks that one slot names the setting and three slots don't. I updated the old assertions that expected the previous sentence ending, in test/vitals.test.ts and test/red-main.test.ts.
  • Left alone: the sentinel.test.ts fixture row that holds the old sentence is stored input, not an assertion on the new text.
  • If it fails again: the next failure's log should show whether that test is flaky on its own or this change really interacts with it.
commented

Parked by the crew: the check failed ("a quiet writer costs almost no CPU while the reader waits") and the decider saw no point in another try

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs