Bug #53129: BlueFS truncate() and poweroff can create corrupted files - bluestore - Ceph

Actions

Copy link

Bug #53129

closed

BlueFS truncate() and poweroff can create corrupted files

Added by Adam Kupczyk over 2 years ago. Updated almost 2 years ago.

Status:

Resolved

Priority:

Normal

Assignee:

Adam Kupczyk

Target version:

% Done:

Source:

Tags:

Backport:

pacific, octopus, nautilus

Regression:

Yes

Severity:

3 - minor

Reviewed:

Affected Versions:

ceph-qa-suite:

Pull request ID:

43774

Crash signature (v1):

Crash signature (v2):

Description

It is possible to create condition in which a BlueFS contains file that is corrupted.
It can happen when BlueFS replay log is on device A and we just wrote to device B and truncated file.

Scenario:
1) write to file h1 on SLOW device
2) flush h1 (initiate data transfer, but no fdatasync yet)
3) truncate h1
4) write to file h2 on DB
5) fsync h2 (forces replay log to be written, after fdatasync to DB)
6) poweroff

In result we have file h1 that is properly declared in replay log, but with uninitialized content.
This happens even with
https://tracker.ceph.com/issues/50965
applied.

I think it is regression introduced by above fix.

Related issues 3 (0 open — 3 closed)