Issue evidence · Run · nextflow-io/nextflow
GitHub issue: https://github.com/nextflow-io/nextflow/issues/7816
Not reproduced
A race on availCpus/availMemory in LocalPollingMonitor sometimes leaves a pending task unsubmitted so the run never ends.
Ran a stub workflow of 46 task attempts (6 exit 137 on the first attempt and retry, 40 instant tasks) with executor.cpus=16 and executor.memory=64 GB, 24 times in a row with Nextflow 26.04.7 on macOS (arm64, Temurin 21), each with a 150 s cap.
The reporter saw 22 hangs in 600 runs (about 4%), so 24 runs would show none about 40% of the time: this does not show the race is absent. Their reproduction is a 4-CPU CI runner and was not run.
Checked 2026-10-14 in Badgr’s local development environment.
← All issue evidence