Project

General

Profile

Feature #2911

osd: Restrict recovery when the OSD full list is nonempty

Added by Greg Farnum about 8 years ago. Updated about 8 years ago.

Status:
Duplicate
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Reviewed:
Affected Versions:
Pull request ID:

Description

See the conversation at http://www.spinics.net/lists/ceph-devel/msg08010.html

It would be nice if we could somehow restrict recovery when an OSD is full, to prevent sequential OSD failures as degraded PGs get replicated onto nodes, fill them, and cause crashes. We can certainly (as an option) prevent the monitor from marking OSDs out whenever the full list is non-empty, but we should think about if there are other mitigating actions we can take.


Related issues

Related to Ceph - Feature #15910: Increase the default value of mon_osd_min_in_ratio Resolved 05/17/2016
Duplicates Ceph - Feature #1637: OSDs running full take down other OSDs Duplicate 10/20/2011

History

#1 Updated by Sage Weil about 8 years ago

  • translation missing: en.field_position set to 42

#2 Updated by Sage Weil about 8 years ago

  • Subject changed from Restrict recovery when the OSD full list is nonempty to osd: Restrict recovery when the OSD full list is nonempty
  • translation missing: en.field_position deleted (42)
  • translation missing: en.field_position set to 42

#3 Updated by Sage Weil about 8 years ago

  • translation missing: en.field_position deleted (42)
  • translation missing: en.field_position set to 2

#4 Updated by Sage Weil about 8 years ago

  • translation missing: en.field_story_points set to 5
  • translation missing: en.field_position deleted (5)
  • translation missing: en.field_position set to 3

#5 Updated by Sage Weil about 8 years ago

  • Status changed from New to Duplicate

#6 Updated by David Zafman over 4 years ago

  • Related to Feature #15910: Increase the default value of mon_osd_min_in_ratio added

Also available in: Atom PDF