Watchdog caught collective operation timeout WorkNCCL AllGather
NCCL collective timed out. One rank stopped participating. This entry explains how to confirm the cause, apply the fix, and separate it from adjacent pytorch-fsdp-ddp failures.
Watchdog caught collective operation timeout WorkNCCL AllGather means NCCL collective timed out. One rank stopped participating. Preserve the first preceding error, then run the targeted control below.
- Symptom
Watchdog caught collective operation timeout WorkNCCL AllGather- Root cause
- NCCL collective timed out. One rank stopped participating. The decisive evidence is the first log line that precedes "Watchdog caught collective operation timeout WorkNCCL AllGather" and differs from a healthy run.
- Recommended fix
NCCL_IB_TIMEOUT=22- How Denpex helps
- Denpex investigates Watchdog caught collective operation timeout WorkNCCL AllGather using the evidence you provide or your connected workload collects. Earlier rank, host or application evidence is needed to distinguish an initiating failure from a downstream report.
What this failure is
The literal signature is "Watchdog caught collective operation timeout WorkNCCL AllGather". It is a distributed training failure associated with PyTorch Distributed, FSDP, DDP, and TorchElastic. The line identifies the failing operation or subsystem, while the surrounding evidence decides whether it is the initiating fault or a downstream symptom.
Is this what broke your run? Paste your log.
You're reading about Watchdog caught collective operation timeout WorkNCCL AllGather. Paste your own traceback and relevant evidence for an investigation of your workload, with a next action or a specific missing fact. A reference entry does not establish your cause. No account or card for the free diagnosis. Review data handling before submitting sensitive logs.
Before uploading, review cloud data handling and local options.
Why it happens (the mechanism)
NCCL collective timed out. One rank stopped participating. The failure becomes visible at this call site because the operation first requires the missing resource, valid state, healthy peer, or correct result. Earlier log lines and a known-good control carry more causal value than the final wrapper exception.
What you'll observe
- The workload stops or loses forward progress after emitting "Watchdog caught collective operation timeout WorkNCCL AllGather".
- A retry on the same configuration reproduces the failure because the causal state has not changed.
- The outer framework exception can hide the rank, node, allocation, or dependency that failed first.
- Increasing timeouts or reducing workload size can suppress the symptom without correcting the cause.
Common symptoms and what they mean
| Symptom | Why it happens |
|---|---|
| Watchdog caught collective operation timeout WorkNCCL AllGather | NCCL collective timed out. One rank stopped participating. |
| The same operation fails at a consistent stage of PyTorch Distributed, FSDP, DDP, and TorchElastic. | The decisive evidence is the first log line that precedes "Watchdog caught collective operation timeout WorkNCCL AllGather" and differs from a healthy run. |
| The first related warning appears before the final exception and names the causal subsystem. | A nearby failure remains a competing hypothesis until a control separates configuration, capacity, transport, and hardware causes. |
| A known-good control changes one variable and either reproduces or clears the failure. | NCCL collective timed out. One rank stopped participating. |
Which systems are affected
- PyTorch Distributed, FSDP, DDP, and TorchElastic
- production-shaped multi-accelerator workloads
- containerized and bare-metal deployments of the same stack
How to confirm this is the problem
Use this checklist to test the hypothesis against a small reproduction. No single line proves the root cause, so preserve the preceding events and compare one variable at a time.
- ✓Find the first occurrence of "Watchdog caught collective operation timeout WorkNCCL AllGather" and preserve at least 100 lines before it.
- ✓Identify which rank, node, device, or process emitted the first related warning.
- ✓Compare the failing rank, node, input, or configuration with one known-good control.
- ✓Repeat the same input after the targeted change and require the signature to disappear.
- ✓Resume from latest checkpoint only after the control passes.
Root cause
- NCCL collective timed out. One rank stopped participating.
- The decisive evidence is the first log line that precedes "Watchdog caught collective operation timeout WorkNCCL AllGather" and differs from a healthy run.
- A nearby failure remains a competing hypothesis until a control separates configuration, capacity, transport, and hardware causes.
The fix and how to prevent it
Searchable error signature
Watchdog caught collective operation timeout WorkNCCL AllGatherUse this text as a lookup key in logs and upstream issue trackers. It is not presented as a captured customer log. Confirm the cause from your own preceding events, versions, configuration and the cited references.
The fix and the prevention pattern
The root cause is on this page and stays free. A free account adds the exact remediation steps, saved history, and the fix on every entry in the encyclopedia.
Sign up free. Unlock the full analysisNo credit card. Daily allowance follows verified trust tier. Instant access.
Why the recommended fix works
Check the failing rank with nvidia-smi and kill zombie processes. Raise the collective timeout via init_process_group(timeout=timedelta(minutes=30)) (there is no NCCL_TIMEOUT env var); for watchdog tuning use TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC. For IB-level retries: NCCL_IB_TIMEOUT=22 (valid range 0-31). This changes the condition in the causal diagnosis instead of hiding the outer exception. The repeated control proves ownership before recovery from latest checkpoint.
Code examples
# Preserve evidence before restarting
rg -n -i 'error|exception|timeout|failed' <log-file>
nvidia-smi
python -m torch.utils.collect_env
# Find the exact signature in the complete log
rg -n -F -- "Watchdog caught collective operation timeout WorkNCCL AllGather" <log-file>Adapt the snippet to your framework. The same pattern holds for PyTorch Lightning, Hugging Face Trainer, DeepSpeed, Megatron-LM, and vLLM training wrappers. Where the wrapper exposes a config flag (for examplelr_scheduler_type in Trainer), prefer the flag over the imperative API to keep the schedule declarative and reproducible.
Best practices by model family
| Model / Stack | Recommendation | Notes |
|---|---|---|
| First response | Preserve the first failure | Keep the context before "Watchdog caught collective operation timeout WorkNCCL AllGather" so aggregation does not erase causality. |
| Confirmation | Change one variable | Use a known-good node, rank, input, or configuration as the control. |
| Recovery | Resume from latest checkpoint | Resume only after the literal signature no longer appears in the same control. |
With the fix vs without the fix
| Dimension | With the fix | Without the fix |
|---|---|---|
| Evidence | First preceding error and one controlled comparison | Only the final aggregated exception |
| Fix | Check the failing rank with nvidia-smi and kill zombie processes. Raise the collective timeout via init_process_group(timeout=timedelta(minutes=30)) (there is no NCCL_TIMEOUT env var); for watchdog tuning use TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC. For IB-level retries: NCCL_IB_TIMEOUT=22 (valid range 0-31). | Retrying the unchanged workload |
| Exit criterion | "Watchdog caught collective operation timeout WorkNCCL AllGather" is absent in the repeated control | The job happened to run once |
Diagnostic note
“Treat "Watchdog caught collective operation timeout WorkNCCL AllGather" as a search key and an investigation checkpoint, not as proof of every cause associated with the phrase. The high-value evidence is what changed immediately before it and whether the failure follows the workload, node, or configuration.”
Visual fingerprint
literal error captured
|
v
find first preceding failure
|
v
run one known-good control
|
+-- follows workload --> inspect input or configuration
+-- follows node ------> inspect hardware or platform
+-- disappears --------> validate the targeted fixDiagnose this failure in VS Code
Select the traceback or open the failed terminal, then run Denpex locally to see the initiating rank, collateral failures, exact fix, and verification command without uploading the log.
Install the free VS Code extensionNCCL errors in context
NCCL is where a distributed job reports failure, which is not the same as where it failed. The hub lists every common NCCL error next to what it actually indicates, and the environment variables that tell them apart.
Compare every nccl error side by sideRelated failures to investigate next
Frequently asked questions
Questions engineers and on-call staff commonly ask about this failure.
What does "Watchdog caught collective operation timeout WorkNCCL AllGather" mean?
Is this line always the root cause?
What should I collect before restarting?
What is the fastest confirmation?
How do I prevent it from recurring?
Don't just read the fix, diagnose your run
The encyclopedia tells you what went wrong. Denpex tells you what went wrong in YOUR training run. With your logs, your config, and your stack.