Skip to content

Checkpoint I/O Stall (NFS RPC Saturation)

Checkpoint writes stall because the NFS RPC layer saturates its slot table, producing a 'bandwidth paradox' where the network is barely utilized yet I/O is the bottleneck.

Quick answer

Checkpoint writes stall because the NFS RPC layer saturates its slot table, producing a 'bandwidth paradox' where the network is barely utilized yet I/O is the bottleneck.

Infrastructure#checkpoint#nfs#io-stall#rpc#infrastructure#storage

What this failure is

Checkpoint I/O Stall (NFS RPC Saturation) is a Infrastructure failure seen during ML training runs. Checkpoint writes stall because the NFS RPC layer saturates its slot table, producing a 'bandwidth paradox' where the network is barely utilized yet I/O is the bottleneck. Common tags: Checkpoint, Nfs, Io Stall, Rpc.

Live diagnosis, no signup

Is this what broke your run? Paste your log.

You're reading about Checkpoint I/O Stall (NFS RPC Saturation). 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.

training_logs.txt
No log to hand? Try one:

3 free diagnoses/day

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.

Evaluate one incident

Why it happens (the mechanism)

NFS RPC layer saturates its fixed slot table before bandwidth is used. Many ranks issue small synchronous writes at once. Metadata/RPC overhead dominates, not throughput. 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

  • Checkpoint save takes far longer than expected
  • Network shows low utilization yet I/O is the bottleneck
  • All ranks block on the checkpoint write

Common symptoms and what they mean

SymptomWhy it happens
NFS RPC slot table saturated (e.g. 128 slots full)NFS RPC layer saturates its fixed slot table before bandwidth is used
1-10% utilization of a 200 Gbps link during save (bandwidth paradox)Many ranks issue small synchronous writes at once
Long checkpoint write duration / stallMetadata/RPC overhead dominates, not throughput

Which systems are affected

  • Clusters checkpointing to shared NFS
  • Synchronous full-model checkpoints
  • High rank counts writing simultaneously

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: NFS RPC slot table saturated (e.g. 128 slots full)
  • ✓Verified signal present: 1-10% utilization of a 200 Gbps link during save (bandwidth paradox)
  • ✓Verified signal present: Long checkpoint write duration / stall
  • ✓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 analysis

No 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 extension

Related failures to investigate next

Lambda Bgp Route Withdrawal Region Outage often co-occurs with this failure because both share underlying assumptions about infrastructure; if you fix this entry but still see drift, validate the related primary cause next. lambda bgp route withdrawal region outage →
Kubernetes Gpu Pod Pending often co-occurs with this failure because both share underlying assumptions about infrastructure; if you fix this entry but still see drift, validate the related neighboring cause next. kubernetes gpu pod pending →
Cloud Storage Throttling often co-occurs with this failure because both share underlying assumptions about infrastructure; if you fix this entry but still see drift, validate the related neighboring cause next. cloud storage throttling →

Root cause

  • NFS RPC layer saturates its fixed slot table before bandwidth is used
  • Many ranks issue small synchronous writes at once
  • Metadata/RPC overhead dominates, not throughput

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.

We send a single-use code tied to that address. Static provider and TLD rules do not reject valid addresses. Account trust determines the benefit after signup.

References

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.

Diagnose your logs freeBrowse all failures

Related Infrastructure errors

Dual ISP BGP Route Withdrawal Causing Complete GPU Cloud Region Outage

Infrastructure · critical

Routine UPS Maintenance Triggering Cascading Power and Cooling Failure Across GPU Cloud Region

Infrastructure · critical

Remediation Storm Prevention via Circuit Breaker Pattern in AutoClusters

Infrastructure · high

Network Storage Volume Causing Process Hangs on H100 GPU Nodes

Infrastructure · high