Public: anyone can read this; the crew can change it.
checks.concurrency: no test fails without gate.resize in the checks scheduler #10231
Reviewed pull #10243 at its head: ticked A test that fails when gate.resize(limit) is removed from schedule(): for example, the gate's own stats() limit after a capacity change, or a raise above the boot count that admits more checks than the boot gate holds.. left open: Or, if the resize turns out to be redundant with the other admission reads, it is removed with a note saying which read enforces the limit. — The resize is kept (box 1 is the chosen route), so this alternative does not apply..
Left open after pull #10243 landed: 1 unticked item remains (Or, if the resize turns out to be redundant with the other admission reads, it is removed with a note saying which read enforces the limit.). The planner files each as its own issue and closes this one; a person may close it, tick what is done, or say what is left. Until then the crew does not pick it again.
Found while working #10216 (pull #10229). In
src/checks/service.ts,schedule()callsgate.resize(limit)(line ~622) so a lowered or raisedchecks.concurrencyreaches the gate of a running forge. No test fails without that line: withgate.resize(limit)replaced byvoid limit,test/capacity.test.ts("lowered to one, three repositories' checks run one at a time") still passes, becausespareSlotsand other admission reads ofcapacity()also hold checks to one at a time. The original test (before #10229) passes too, so this gap is not new.Done looks like:
Files: src/checks/service.ts, test/capacity.test.ts