Project

General

Profile

Actions

Feature #2651

closed

mon: race calling tick() when doing slurping

Added by Greg Farnum almost 12 years ago. Updated about 11 years ago.

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

0%

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

Description

Right now the monitor calls tick() on all the PaxosService implementations when it's doing slurping. This introduces a race, in the case that there is slurping going on which has changed the on-disk data without yet putting it back into a consistent state.

There is a quick fix in wip-tick-slurping-race, but we should come up with a stronger solution (integrate slurping into the Paxos state system, probably?) for master. (Or, hopefully, merge Joao's single-paxos stuff soon enough that we don't need to worry about it.)


Related issues 1 (0 open1 closed)

Has duplicate Ceph - Bug #2646: mon:update_from_paxos: error parsing incremental update: buffer::end_of_bufferResolved06/22/2012

Actions
Actions #1

Updated by Sage Weil almost 12 years ago

  • Status changed from New to Resolved
Actions #2

Updated by Sage Weil almost 12 years ago

  • Status changed from Resolved to 12

oops, stronger fix, yes!

Actions #3

Updated by Sage Weil almost 12 years ago

  • Tracker changed from Bug to Cleanup

making this a cleanup so that it stops confusing me :)

Actions #4

Updated by Sage Weil about 11 years ago

  • Tracker changed from Cleanup to Feature
  • Status changed from 12 to Rejected

no more slurp.

Actions

Also available in: Atom PDF