Project

General

Profile

Actions

Backport #39501

closed

mimic: snapshot object maps can go inconsistent during copyup

Added by Nathan Cutler almost 5 years ago. Updated about 3 years ago.

Status:
Rejected
Priority:
Normal
Assignee:
Jason Dillaman
Target version:
-
Release:
mimic
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Related issues 2 (0 open2 closed)

Follows rbd - Backport #39196: mimic: Several race conditions are possible between io::ObjectRequest and io::CopyupRequestRejectedJason DillamanActions
Copied from rbd - Bug #39435: snapshot object maps can go inconsistent during copyupResolvedIlya Dryomov

Actions
Actions #1

Updated by Nathan Cutler almost 5 years ago

  • Copied from Bug #39435: snapshot object maps can go inconsistent during copyup added
Actions #2

Updated by Nathan Cutler almost 5 years ago

  • Follows Backport #39196: mimic: Several race conditions are possible between io::ObjectRequest and io::CopyupRequest added
Actions #3

Updated by Nathan Cutler almost 5 years ago

  • Status changed from New to Need More Info
  • Assignee set to Nathan Cutler
Actions #4

Updated by Jason Dillaman over 4 years ago

  • Status changed from Need More Info to In Progress
  • Assignee changed from Nathan Cutler to Jason Dillaman
Actions #5

Updated by Oliver Freyermuth about 4 years ago

This is probably the reason for the corruption observed when performing backups based on diff output when fast-diff is active:
https://github.com/elemental-lf/benji/issues/53
While we luckily did not see corruption, we see invalid object maps on a daily basis (probably since we use fstrim) and repair them.

Since we'll stay on Mimic still for our virtualization cluster for a while, I'd love to see this backport happen.

Actions #6

Updated by Nathan Cutler about 3 years ago

  • Status changed from In Progress to Rejected

mimic EOL

Actions

Also available in: Atom PDF