slurmstepd: task/cgroup: unable to allocate requested memory for GPU step

slurmstepd could not create the memory cgroup for the step. The requested amount exceeds what the node can back, or the cgroup hierarchy is misconfigured, the job never starts rather than being killed later. This entry explains how to confirm the cause, apply the fix, and separate it from adjacent slurm failures.

The literal signature is "slurmstepd: task/cgroup: unable to allocate requested memory for GPU step". It is a infrastructure failure associated with Slurm GPU clusters and cgroup-constrained jobs. The line identifies the failing operation or subsystem, while the surrounding evidence decides whether it is the initiating fault or a downstream symptom.

slurmstepd could not create the memory cgroup for the step. The requested amount exceeds what the node can back, or the cgroup hierarchy is misconfigured, the job never starts rather than being killed later. 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 "slurmstepd: task/cgroup: unable to allocate requested memory for GPU step".
• 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.

• Slurm GPU clusters and cgroup-constrained jobs
• production-shaped multi-accelerator workloads
• containerized and bare-metal deployments of the same stack

Apply the following checklist to a small reproduction: each box below is a positive signal that you are looking at this exact failure rather than a sibling in the same taxonomy.

• ✓Find the first occurrence of "slurmstepd: task/cgroup: unable to allocate requested memory for GPU step" and preserve at least 100 lines before it.
• ✓Identify which rank, node, device, or process emitted the first related warning.
• ✓check cgroup.conf (ConstrainRAMSpace, AllowedRAMSpace) and confirm the cgroup version matches what this Slurm build expects, a node migrated to cgroup v2 with a Slurm built for v1 fails exactly here.
• ✓Repeat the same input after the targeted change and require the signature to disappear.
• ✓Resume from job resubmission with a corrected memory request only after the control passes.

slurmstepd: task/cgroup: unable to allocate requested memory for GPU step

Timestamps and exact values vary across runs, but the pattern. An info-level start, an early WARN, an ERROR carrying the symptom. Is the actual fingerprint you should alert on. The Denpex platform flags this combination automatically.

compare the request against the node, scontrol show node <node> | grep RealMemory versus the --mem or --mem-per-gpu on the job. A request above RealMemory can be accepted by the scheduler and then fail at step creation. This changes the condition in the causal diagnosis instead of hiding the outer exception. The repeated control proves ownership before recovery from job resubmission with a corrected memory request.

# Preserve evidence before restarting
scontrol show job <job-id>
sacct -j <job-id> --format=JobID,State,ExitCode,MaxRSS

# Find the exact signature in the complete log
rg -n -F -- "slurmstepd: task/cgroup: unable to allocate requested memory for GPU step" <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.

• slurmstepd could not create the memory cgroup for the step. The requested amount exceeds what the node can back, or the cgroup hierarchy is misconfigured, the job never starts rather than being killed later.
• The decisive evidence is the first log line that precedes "slurmstepd: task/cgroup: unable to allocate requested memory for GPU step" and differs from a healthy run.
• A nearby failure remains a competing hypothesis until a control separates configuration, capacity, transport, and hardware causes.