Project

General

Profile

Feature #1912

mds: should time out slow monitors

Added by Greg Farnum almost 12 years ago. Updated about 7 years ago.

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

0%

Source:
Tags:
Backport:
Reviewed:
Affected Versions:
Component(FS):
Labels (FS):
Pull request ID:

Description

Similar to #1841 for the OSDs, the MDS should reconnect to a different monitor if its beacons go too long without getting acked. This will prevent unfortunate scenarios where a standby MDS doesn't discover it needs to go active until the messenger timeout is exceeded! (See Karoly Horvath's email thread "how can I achieve HA with ceph?".)

History

#1 Updated by Sage Weil almost 12 years ago

  • Subject changed from The MDS should time out slow monitors to mds: should time out slow monitors
  • translation missing: en.field_position set to 30

#2 Updated by Sage Weil almost 12 years ago

  • translation missing: en.field_story_points set to 2
  • translation missing: en.field_position deleted (30)
  • translation missing: en.field_position set to 30

#3 Updated by Sage Weil almost 12 years ago

  • translation missing: en.field_position deleted (30)
  • translation missing: en.field_position set to 2

#4 Updated by Greg Farnum almost 12 years ago

  • Status changed from New to Resolved

This was actually very easy to do. Ripping out the utterly useless beacon_kill code was actually a larger diff!

#5 Updated by Greg Farnum almost 12 years ago

Errm, commit:ac9b2d092f1f075621f30f77cdbb49bdf35b9ae5

#6 Updated by John Spray about 7 years ago

  • Project changed from Ceph to CephFS
  • Category deleted (1)
  • Target version deleted (v0.41)

Bulk updating project=ceph category=mds bugs so that I can remove the MDS category from the Ceph project to avoid confusion.

Also available in: Atom PDF