Project

General

Profile

Bug #43838

cephadm: Forcefully Remove Services (unresponsive hosts)

Added by Sebastian Wagner 6 months ago. Updated about 2 months ago.

Status:
In Progress
Priority:
High
Category:
cephadm/scheduler
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature:

Description

If a host is no longer responsive, users need a way to remove services on this host from ceph, despite the host no longer being there.

  • How to deal with hanging I/O operations, or when the OSD disk fails to unmount?
  • This also contains a check for existing daemons that are running there

Related issues

Related to Orchestrator - Bug #44252: cephadm: mgr,mds scale-down should prefer standby daemons New
Related to Orchestrator - Tasks #45143: cephadm: scheduler improvements Closed

History

#1 Updated by Sage Weil 5 months ago

One option is to have them 'ceph orch host rm $hostname'...

#2 Updated by Sebastian Wagner 3 months ago

  • Priority changed from Normal to High

#3 Updated by Sebastian Wagner 3 months ago

  • Description updated (diff)

#4 Updated by Sebastian Wagner 3 months ago

  • Status changed from New to In Progress
  • Assignee set to Sebastian Wagner

#5 Updated by Sebastian Wagner 3 months ago

  • Related to Bug #44252: cephadm: mgr,mds scale-down should prefer standby daemons added

#6 Updated by Sebastian Wagner 3 months ago

  • Related to Tasks #45143: cephadm: scheduler improvements added

#7 Updated by Sebastian Wagner about 2 months ago

  • Category changed from cephadm to cephadm/scheduler

Also available in: Atom PDF