Project

General

Profile

Actions

Bug #55779

open

fuse client losing connection to mds

Added by Milind Changire almost 2 years ago. Updated 7 months ago.

Status:
Need More Info
Priority:
Normal
Category:
Correctness/Safety
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Component(FS):
ceph-fuse
Labels (FS):
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

fuse client losing connection to mds after about 5 minutes of I/O

  • no stack trace in mds logs
  • no mds crashes
Actions #1

Updated by Venky Shankar almost 2 years ago

  • Status changed from New to Need More Info

The client might be getting blocklisted. Could you share the client/mds logs?

Actions #2

Updated by Milind Changire almost 2 years ago

Venky Shankar wrote:

The client might be getting blocklisted. Could you share the client/mds logs?

Here are the logs:
ceph-post-file: 35edb9f5-6e3b-4e0c-9dac-733dc0963527

Actions #3

Updated by Venky Shankar almost 2 years ago

The MDS got a termination signal. E.g.: mds.a:

2022-05-29T22:18:09.220+0530 7ff3eae14640 -1 received  signal: Terminated from  (PID: 70532) UID: 25405
2022-05-29T22:18:09.220+0530 7ff3eae14640 -1 mds.a *** got signal Terminated ***
2022-05-29T22:18:09.220+0530 7ff3eae14640  1 mds.a suicide! Wanted state up:active
2022-05-29T22:18:09.220+0530 7ff3eae14640  5 mds.beacon.a set_want_state: up:active -> down:dne
2022-05-29T22:18:09.220+0530 7ff3eae14640  5 mds.beacon.a Sending beacon down:dne seq 88

Who is PID 70532?

Actions #4

Updated by Venky Shankar almost 2 years ago

  • Category set to Correctness/Safety
  • Assignee set to Kotresh Hiremath Ravishankar
  • Target version set to v18.0.0

Kotresh, please take a look at this. Milind mentioned that this is easily reproducible.

Actions #6

Updated by Patrick Donnelly 7 months ago

  • Target version deleted (v18.0.0)
Actions

Also available in: Atom PDF