Project

General

Profile

Actions
Copy link

Bug #42056

open

rgw: librgw write wrongly closed in NFS3

Added by Chenjiong Deng over 4 years ago. Updated over 4 years ago.

Status:
New
Priority:
Normal
Assignee:
Matt Benjamin
Target version:
Ceph - v12.2.13
% Done:

0%

Source:
Tags:
Backport:
nautilus, mimic, luminous
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

We mount nfs via librgw, then dd test on export directory.

[root@dev]# mount -t nfs -o noauto,noacl,soft,nfsvers=3,proto=tcp,nolock,sync xxx:/ /mnt

But dd has a write error：

You can see that the offset is not ordered：

Both the dd sequential write and the FIO sequential write can reproduce the nfs3_write problem.
When bs is no more than 1MB, dd and FIO have no recurring problems after multiple tests, and bs>=2MB will reproduce the problem.

Files

Download all files
dd.jpeg (54.2 KB) dd.jpeg Chenjiong Deng, 09/26/2019 07:11 AM
offseterror.jpeg (339 KB) offseterror.jpeg Chenjiong Deng, 09/26/2019 07:16 AM
Actions
Copy link
 #1

Updated by Chenjiong Deng over 4 years ago

Based on tests, Luminous,Mimic,Nautilus，they all have this problem.

Actions
Copy link
 #2

Updated by Nathan Cutler over 4 years ago

  • Backport set to nautilus, mimic, luminous
Actions
Copy link
 #3

Updated by Casey Bodley over 4 years ago

  • Assignee set to Matt Benjamin
Actions
Copy link

Also available in: Atom PDF