DeepSpeed PPO Actor forward pass deadlocked waiting for Critic ZeRO-3 parameter gather

The actor blocked waiting for a ZeRO-3 parameter gather owned by the critic. Two ZeRO-3 models in one process share a parameter coordinator; if their forward passes interleave, each can wait on a collective the other has not entered. This entry explains how to confirm the cause, apply the fix, and separate it from adjacent rlhf failures.

Symptom

DeepSpeed PPO Actor forward pass deadlocked waiting for Critic ZeRO-3 parameter gather

Root cause

The actor blocked waiting for a ZeRO-3 parameter gather owned by the critic. Two ZeRO-3 models in one process share a parameter coordinator; if their forward passes interleave, each can wait on a collective the other has not entered. The decisive evidence is the first log line that precedes "DeepSpeed PPO Actor forward pass deadlocked waiting for Critic ZeRO-3 parameter gather" and differs from a healthy run.

Recommended fix

serialise the two models' forward passes, never run actor and critic concurrently in the same process under ZeRO-3.

How Denpex helps

Denpex investigates DeepSpeed PPO Actor forward pass deadlocked waiting for Critic ZeRO-3 parameter gather 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.

The literal signature is "DeepSpeed PPO Actor forward pass deadlocked waiting for Critic ZeRO-3 parameter gather". It is a distributed training failure associated with TRL, OpenRLHF, Ray, and actor-critic training. The line identifies the failing operation or subsystem, while the surrounding evidence decides whether it is the initiating fault or a downstream symptom.

The actor blocked waiting for a ZeRO-3 parameter gather owned by the critic. Two ZeRO-3 models in one process share a parameter coordinator; if their forward passes interleave, each can wait on a collective the other has not entered. 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.

• The workload stops or loses forward progress after emitting "DeepSpeed PPO Actor forward pass deadlocked waiting for Critic ZeRO-3 parameter gather".
• 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.

• TRL, OpenRLHF, Ray, and actor-critic training
• production-shaped multi-accelerator workloads
• containerized and bare-metal deployments of the same stack

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 "DeepSpeed PPO Actor forward pass deadlocked waiting for Critic ZeRO-3 parameter gather" and preserve at least 100 lines before it.
• ✓Identify which rank, node, device, or process emitted the first related warning.
• ✓py-spy dump on the hung rank shows which gather each model is inside. If both are in a parameter gather, this is the deadlock; if only one is, the other rank has already died and this is collateral.
• ✓Repeat the same input after the targeted change and require the signature to disappear.
• ✓Resume from latest checkpoint only after the control passes.

• The actor blocked waiting for a ZeRO-3 parameter gather owned by the critic. Two ZeRO-3 models in one process share a parameter coordinator; if their forward passes interleave, each can wait on a collective the other has not entered.
• The decisive evidence is the first log line that precedes "DeepSpeed PPO Actor forward pass deadlocked waiting for Critic ZeRO-3 parameter gather" and differs from a healthy run.
• A nearby failure remains a competing hypothesis until a control separates configuration, capacity, transport, and hardware causes.

DeepSpeed PPO Actor forward pass deadlocked waiting for Critic ZeRO-3 parameter gather

Use 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.

serialise the two models' forward passes, never run actor and critic concurrently in the same process under ZeRO-3. This changes the condition in the causal diagnosis instead of hiding the outer exception. The repeated control proves ownership before recovery from latest checkpoint.

# 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 -- "DeepSpeed PPO Actor forward pass deadlocked waiting for Critic ZeRO-3 parameter gather" <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.

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.

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.