Skip to content

#253: fix provider-workload gate ENOENT race (scope_failed flake) #254

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
seungpyoson merged 3 commits into from
Jun 26, 2026
Merged

#253: fix provider-workload gate ENOENT race (scope_failed flake) #254

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
d8766e3
fix(review-workload): retry provider gate ENOENT lost mid-acquire (#253)
Jun 26, 2026
0f4ebfe
fix(review-workload): re-establish vanished lock root instead of spin…
Jun 26, 2026
125cc9f
Merge main into fix/253-grok-workload-gate-race (bring current)
Jun 26, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 14 additions & 2 deletions plugins/agy/scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
16 changes: 14 additions & 2 deletions plugins/api-reviewers/scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
16 changes: 14 additions & 2 deletions plugins/claude/scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
16 changes: 14 additions & 2 deletions plugins/gemini/scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
16 changes: 14 additions & 2 deletions plugins/grok/scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
16 changes: 14 additions & 2 deletions plugins/kimi/scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
16 changes: 14 additions & 2 deletions relay/relay-agy/scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
16 changes: 14 additions & 2 deletions relay/relay-gemini/scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
16 changes: 14 additions & 2 deletions relay/relay-grok/scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
16 changes: 14 additions & 2 deletions relay/relay-kimi/scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
16 changes: 14 additions & 2 deletions scripts/lib/review-workload.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -211,14 +211,26 @@ function acquireProviderWorkloadGate(file, env) {
release: () => releaseProviderWorkloadGate(gateDir, token),
});
} catch (error) {
if (error?.code !== "EEXIST") throw error;
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
if (error?.code !== "EEXIST" && error?.code !== "ENOENT") throw error;
if (Date.now() >= deadline) {
return Object.freeze({
ok: false,
holder: parseGateOwner(readGateOwnerRaw(gateDir)),
});
}
// ENOENT from mkdirSync itself means the lock root (gateDir's parent) vanished
// under us — a concurrent teardown, not the gate-dir mid-acquire race. Reclaim
// would loop with no progress (renameSync also ENOENTs → true → continue), so
// re-establish the parent and sleep-pace: a persistently-missing root degrades
// to a bounded poll, never a CPU hot-loop. The race ENOENT instead comes from
// the owner-file open (syscall !== "mkdir") and falls through to reclaim, where
// the next mkdir succeeds.
if (error?.code === "ENOENT" && error?.syscall === "mkdir") {
try { mkdirSync(lockRoot(env), { recursive: true, mode: 0o700 }); } catch { /* best effort */ }
sleepSync(GATE_POLL_MS);
continue;
}
if (tryReclaimProviderWorkloadGate(gateDir, env)) continue;
sleepSync(GATE_POLL_MS);
}
Comment thread
seungpyoson marked this conversation as resolved.
}
Expand Down
Loading
Loading