stablyai/orca · error · Error

Benchmark process group changed before signaling

Error message

Benchmark process group changed before signaling

What it means

Thrown after SIGSTOP'ing the whole process group (-pgid). The code re-enumerates members of pgid and requires (a) at least one member survives and (b) every surviving member's command still contains environmentFragment. Failure means the group's composition drifted during the stop — members exited or were replaced by unrelated processes sharing the pgid.

Source

Thrown at config/scripts/macos-computer-helper-owner-loss-processes.mjs:295

    operations.signalProcess(anchor.pid, 'SIGSTOP')
    groupState.anchorPid = anchor.pid
    const stoppedAnchor = operations
      .processIdentities(true)
      .find((identity) => identity.pid === anchor.pid)
    if (!sameIdentity(stoppedAnchor, anchor)) {
      throw new Error('Benchmark process group anchor changed before signaling')
    }
    operations.signalProcess(-pgid, 'SIGSTOP')
    groupState.stopped = true
    groupState.anchorPid = null
    const stoppedMembers = operations
      .processIdentities(true)
      .filter((identity) => identity.pgid === pgid)
    if (
      stoppedMembers.length === 0 ||
      stoppedMembers.some((identity) => !identity.command.includes(environmentFragment))
    ) {
      throw new Error('Benchmark process group changed before signaling')
    }
    if (signal !== 'SIGSTOP') {
      operations.signalProcess(-pgid, signal)
      if (signal !== 'SIGKILL') {
        operations.signalProcess(-pgid, 'SIGCONT')
      }
      groupState.stopped = false
      groupState.anchorPid = null
    }
    return true
  } catch (error) {
    const recoveryErrors = compensateStoppedGroup(pgid, groupState, operations)
    if (recoveryErrors.length > 0) {
      throw new AggregateError(
        [error, ...recoveryErrors],
        'Benchmark process group signal recovery failed'
      )
    }

View on GitHub (pinned to 1136503c6a)

Solutions

Re-run the trial after re-spawning the helper — group drift is not recoverable mid-signal
Widen environmentFragment to a stable argv token that survives helper refactors (an env var like ORCA_BENCH_ROLE rather than a path)
Spawn the helper with setsid or a dedicated posix_spawnattr pgid to prevent foreign processes joining the group
Log stoppedMembers at the catch site to see whether members vanished or commands drifted

Defensive patterns

Strategy: try-catch

Try / catch

try {
  signalValidatedProcessGroup(pgid, frag, sig, state)
} catch (error) {
  if (/process group changed before signaling/.test(error.message)) {
    respawnHelperAndRetry()
  } else throw error
}

Prevention

Use a stable environmentFragment (an env var token) rather than a brittle path substring
Spawn helpers with setsid or a dedicated pgid so foreign processes cannot join the group
Log stoppedMembers in the catch to distinguish vanished members from command drift

When it happens

Trigger: operations.signalProcess(-pgid, 'SIGSTOP') succeeds, but the subsequent processIdentities(true).filter(pgid === ...) returns either an empty list or members whose command no longer contains environmentFragment.

Common situations: A helper child exits between the anchor stop and the group stop; a setpgid race places an unrelated process into the same pgid; or environmentFragment was chosen too narrowly and no longer matches the helper's actual command line (e.g. after an argv refactor).

Related errors

AI-assisted analysis of stablyai/orca@1136503c6a (2026-08-12). Data as JSON: /api/errors/a93f7588160d2d4b. Report an issue: GitHub.