Project

General

Profile

Actions

Bug #8799

closed

[Node manager] nodes in slurm drained state are counted as "up" but not candidates for shut down

Added by Peter Amstutz about 8 years ago. Updated almost 8 years ago.

Status:
Resolved
Priority:
Normal
Assigned To:
Category:
-
Target version:
Story points:
-

Description

If node manager crashes or otherwise fails during node shutdown, a node can go into drained state; when node manager recovers the drained node is not a candidate for shutdown (because it's not "idle") but is still considered "up".

Node manager should not count these nodes as "up".

In addition, if a node is "drained" but not being actively shut down, node manager should either put it back into idle state, or go ahead and start a new shutdown actor.


Subtasks 2 (0 open2 closed)

Task #8890: Review 8799-make-drained-nodes-idleResolvedPeter Amstutz04/06/2016Actions
Task #8889: Re-enable nodes in drained stateResolvedPeter Amstutz04/06/2016Actions

Related issues

Related to Arvados - Idea #8000: [Node Manager] Shut down nodes in SLURM 'down' stateResolvedActions
Actions

Also available in: Atom PDF