[Crunch] Node went away unexpectedly during pipeline run.
In pipeline 4xphq-d1hrv-rfu9kj37vmke3dh the first two jobs use compute9, then compute9 goes away unexpectedly before the 3rd job can run. After a bit, a new node comes up (compute7) and runs the last job. Explain what happened to compute9.
- Description updated (diff)
Is there a minimum idle time a node must have reached in order to get killed off by node manager during its shutdown window? If not (or if it's short) perhaps that's the explanation/fix we're looking for.
- Category set to Node Manager
- Target version changed from Bug Triage to 2015-04-29 sprint
- Target version changed from 2015-04-29 sprint to Arvados Future Sprints
Also available in: Atom