Feature #5459
open
- Tracker changed from Bug to Feature
- Subject changed from [Crunch] Crunchstat records total load on compute nodes in addition to per-task load to [Crunch] Job logs include node-level resource usage statistics (cpu, ram, swap, etc)
- Category set to Crunch
- Target version changed from Bug Triage to Arvados Future Sprints
This is in addition to (and hopefully in the same format as) the per-task-container stats already provided by crunchstat.
Could be done from crunch-job by running a "crunchstat ... sleep 2^30" process on each node at the beginning of the job, propagating its stats to the log, and killing it off at the end.
- Target version deleted (
Arvados Future Sprints)
- Target version set to Future
Also available in: Atom
PDF
Updated by 
Updated by 
Updated by