Job State Codes

# Job State Codes

## Active States (visible in squeue)

| Code | State | Description |
|------|-------|-------------|
| `PD` | PENDING | Waiting in queue for resources |
| `R` | RUNNING | Executing on compute node(s) |
| `CG` | COMPLETING | Job finishing, cleaning up processes |
| `S` | SUSPENDED | Job suspended (preempted or by admin) |
| `RS` | REQUEUE_HOLD | Held after being requeued |
| `CF` | CONFIGURING | Nodes being configured (booting, etc.) |

---

## Final States (visible in sacct)

| Code | State | Description |
|------|-------|-------------|
| `CD` | COMPLETED | Finished successfully (exit code 0) |
| `F` | FAILED | Finished with non-zero exit code |
| `TO` | TIMEOUT | Killed for exceeding time limit |
| `CA` | CANCELLED | Cancelled by user or admin |
| `OOM` | OUT_OF_MEMORY | Killed for exceeding memory limit |
| `NF` | NODE_FAIL | Failed due to node failure |
| `PR` | PREEMPTED | Preempted by higher-priority job |
| `DL` | DEADLINE | Terminated at deadline |
| `RQ` | REQUEUED | Requeued for re-execution |
| `BF` | BOOT_FAIL | Node failed to boot |
| `SE` | SPECIAL_EXIT | Requeued in held state |
| `RV` | REVOKED | Revoked (federated clusters) |

---

## Exit Codes

Format in sacct: `return_code:signal`

| Example | Meaning |
|---------|---------|
| `0:0` | Success (exit 0, no signal) -- also normal for `TIMEOUT`/`NODE_FAIL`/`CANCELLED`; read the State first |
| `1:0` | Application error (exit 1) |
| `127:0` | Command not found |
| `137:0` | A program inside the script was killed by signal 9 (128+9): OOM reaper, `scancel --signal`, cgroup kill of a step |
| `0:125` | `State=OUT_OF_MEMORY`: the job's cgroup memory limit was hit |
| `0:53` | Batch step could not start (`Reason=RaisedSignal:53(Real-time_signal_19)` on 25.x, `JobLaunchFailure` on older releases) -- `--output`/`--error` directory or `WorkDir` missing on the node; no output file is written |
| `0:15` | The batch script itself received SIGTERM (timeout grace, `scancel`) |
| `0:9` | The batch script itself received SIGKILL (rare; usually a `KillWait` expiry) |

---

## Common Pending Reasons

| Reason | Description | What to Do |
|--------|-------------|------------|
| `(Resources)` | Waiting for nodes/CPUs/memory | Wait, or reduce resource request |
| `(Priority)` | Lower priority than other jobs | Wait (fairshare adjusts) |
| `(Dependency)` | Waiting for dependent job | Check parent job |
| `(DependencyNeverSatisfied)` | Parent job failed | Cancel and resubmit |
| `(QOSMaxJobsPerUserLimit)` | Hit per-user job limit | Wait for jobs to finish |
| `(AssocGrpCPULimit)` | Account CPU limit reached | Wait or check with admin |
| `(AssocGrpGRESLimit)` | Account GRES limit reached | Wait or check with admin |
| `(ReqNodeNotAvail)` | Nodes unavailable | Check `sinfo` for node states |
| `(PartitionTimeLimit)` | Time exceeds partition max | Reduce `--time` |
| `(InvalidAccount)` | Account doesn't exist | Check `sacctmgr show account` |
| `(InvalidQOS)` | QOS not available | Check allowed QOS |
| `(BeginTime)` | Deferred start (`--begin`) | Wait or update |
| `(JobHeldUser)` | Held by user | `scontrol release` |
| `(JobHeldAdmin)` | Held by admin | Contact admin |

Job State Codes

References¶