Project

General

Profile

Bug #46129

mds: fix hang issue when accessing a file under a lost parent directory

Added by Zhi Zhang 5 months ago. Updated about 2 months ago.

Status:
Resolved
Priority:
Normal
Assignee:
Category:
Correctness/Safety
Target version:
% Done:

0%

Source:
Community (dev)
Tags:
Backport:
octopus,nautilus
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Component(FS):
MDS
Labels (FS):
Pull request ID:
Crash signature:

Description

Once a while we had encountered some serious problem that resulted in some metadata lost. After we brought the MDS up, if we looked up a file whose parent directory was not existed any more, this request would hang and other next requests which accessed this lost directory would hang too.

For example, here is a path /dir1/file1. The object of "dir1" in metadata pool was lost after some serious problem.

When doing "ls /dir1", MDS will return EIO as expected.

But before doing "ls /dir", we try doing "ls /dir1/file1" directly, this request will hang forever.


Related issues

Copied to CephFS - Backport #46522: octopus: mds: fix hang issue when accessing a file under a lost parent directory Resolved
Copied to CephFS - Backport #46523: nautilus: mds: fix hang issue when accessing a file under a lost parent directory Resolved

History

#1 Updated by Zhi Zhang 5 months ago

  • Pull request ID set to 35690

#2 Updated by Patrick Donnelly 5 months ago

  • Status changed from New to Fix Under Review
  • Assignee set to Zhi Zhang
  • Target version set to v16.0.0
  • Backport set to octopus,nautilus

#3 Updated by Patrick Donnelly 5 months ago

  • Status changed from Fix Under Review to Pending Backport

#4 Updated by Nathan Cutler 5 months ago

  • Copied to Backport #46522: octopus: mds: fix hang issue when accessing a file under a lost parent directory added

#5 Updated by Nathan Cutler 5 months ago

  • Copied to Backport #46523: nautilus: mds: fix hang issue when accessing a file under a lost parent directory added

#6 Updated by Nathan Cutler about 2 months ago

  • Status changed from Pending Backport to Resolved

While running with --resolve-parent, the script "backport-create-issue" noticed that all backports of this issue are in status "Resolved" or "Rejected".

Also available in: Atom PDF