DeepSpeed ZeRO Stuck / Hang
DeepSpeed ZeRO hangs stall training when gradient synchronization stalls or parameter offload produces deadlocks.
DeepSpeed ZeRO hangs stall training when gradient synchronization stalls or parameter offload produces deadlocks.
What this failure is
DeepSpeed ZeRO Stuck / Hang is a Distributed Training failure seen during ML training runs. DeepSpeed ZeRO hangs stall training when gradient synchronization stalls or parameter offload produces deadlocks. Common tags: Deepspeed, Zero, Hang, Stall.
Is this what broke your run? Paste your log.
You're reading about DeepSpeed ZeRO Stuck / Hang. Paste your own crash log or traceback below and get the real root cause for YOUR run, not this generic entry. No account, no card. Logs are masked at ingress and never saved to account history.
Want 14 days on the Scale plan?
Request an evaluation code. A verified workplace organization activates up to 50 diagnoses a day, alerts, history, and follow-up questions. No credit card or automatic subscription.
Why it happens (the mechanism)
ZeRO stage 3 parameter offload thread deadlocks when CPU memory is exhausted, preventing the GPU from fetching parameters for the next forward pass. Gradient accumulation state corruption: steps_per_print and gradient_accumulation_steps misaligned, causing DeepSpeed's state machine to livelock. NCCL allgather hangs during ZeRO stage 3 parameter gathering when one rank's parameters are not ready (offload not complete). Adam optimizer state update in ZeRO stage 2 stalls if the reduce bucket is not flushed after gradient accumulation steps. Taken together, these mechanisms explain why the failure is reproducible, why it tends to surface on specific workloads or scales, and why generic mitigation attempts often fall short without addressing the underlying cause.
What you'll observe
- Training stops making progress mid-run with no error
- GPUs show near-zero utilization while the job remains alive
- DeepSpeed progress logs stop updating at a specific step
Common symptoms and what they mean
| Symptom | Why it happens |
|---|---|
| RichProgressBar shows no progress for >5 minutes | ZeRO stage 3 parameter offload thread deadlocks when CPU memory is exhausted, preventing the GPU from fetching parameters for the next forward pass |
| nvidia-smi shows 0% GPU utilization on all ranks | Gradient accumulation state corruption: steps_per_print and gradient_accumulation_steps misaligned, causing DeepSpeed's state machine to livelock |
| dmesg shows no errors, process is still running | NCCL allgather hangs during ZeRO stage 3 parameter gathering when one rank's parameters are not ready (offload not complete) |
| DeepSpeed offload thread shows in ps aux but makes no progress | Adam optimizer state update in ZeRO stage 2 stalls if the reduce bucket is not flushed after gradient accumulation steps |
Which systems are affected
- DeepSpeed ZeRO stage 3 with parameter offload to CPU or NVMe
- DeepSpeed with gradient accumulation steps > 1
- Multi-node DeepSpeed training with heterogeneous GPUs
- DeepSpeed with activation checkpointing at high batch sizes
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.
- ✓Reproduce the failure from a clean checkpoint/seed: the symptom must appear without warm-up state from a previous run.
- ✓Verified signal present: RichProgressBar shows no progress for >5 minutes
- ✓Verified signal present: nvidia-smi shows 0% GPU utilization on all ranks
- ✓Verified signal present: dmesg shows no errors, process is still running
- ✓Verified signal present: DeepSpeed offload thread shows in ps aux but makes no progress
- ✓A targeted fix from the "How to fix it" section eliminates or substantially reduces the symptom within one validation pass.
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.
Diagnose 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 extensionDeepSpeed errors in context
DeepSpeed changes when parameters, gradients and optimizer state are created, partitioned, gathered and offloaded. The hub separates ZeRO, memory, checkpoint and pipeline failures by lifecycle phase.
Compare every deepspeed error side by sideRelated failures to investigate next
Root cause
- ZeRO stage 3 parameter offload thread deadlocks when CPU memory is exhausted, preventing the GPU from fetching parameters for the next forward pass
- Gradient accumulation state corruption: steps_per_print and gradient_accumulation_steps misaligned, causing DeepSpeed's state machine to livelock
- NCCL allgather hangs during ZeRO stage 3 parameter gathering when one rank's parameters are not ready (offload not complete)
- Adam optimizer state update in ZeRO stage 2 stalls if the reduce bucket is not flushed after gradient accumulation steps
The fix and how to prevent it
Evaluate Denpex on your own logs
Request a Scale evaluation code. A verified workplace organization activates 14 days with up to 50 diagnoses a day. Every account keeps its current diagnosis allowance and gets a verification path. No card or automatic subscription.
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.