Bug 828039 - ping_pong fails on fuse/nfs mount when new bricks are added to distribute volume
Summary: ping_pong fails on fuse/nfs mount when new bricks are added to distribute volume
Keywords:
Status: CLOSED DEFERRED
Alias: None
Product: GlusterFS
Classification: Community
Component: replicate
Version: 3.3-beta
Hardware: Unspecified
OS: Unspecified
unspecified
high
Target Milestone: ---
Assignee: Pranith Kumar K
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks: 991445
TreeView+ depends on / blocked
 
Reported: 2012-06-04 05:50 UTC by Shwetha Panduranga
Modified: 2014-12-14 19:40 UTC (History)
2 users (show)

Fixed In Version:
Clone Of:
: 991445 (view as bug list)
Environment:
Last Closed: 2014-12-14 19:40:28 UTC
Regression: ---
Mount Type: ---
Documentation: DP
CRM:
Verified Versions:
Embargoed:


Attachments (Terms of Use)
fuse mount log (623.96 KB, text/x-log)
2012-06-04 05:50 UTC, Shwetha Panduranga
no flags Details

Description Shwetha Panduranga 2012-06-04 05:50:54 UTC
Created attachment 589046 [details]
fuse mount log

Description of problem:
-----------------------
When new bricks are added to a plain distribute volume(8 bricks) to change the volume type to distribute-replicate (8x2), ping_pong running on fuse/nfs mount fails. 


Version-Release number of selected component (if applicable):
-----------------------------------------------------------
3.3.0qa45

How reproducible:
------------------
often

Steps to Reproduce:
---------------------
1. Create a plain distribute volume (8 bricks)
2. create fuse/nfs mounts
3. execute "/root/ping_pong_dir/vanilla_ping_pong/ping_pong ./file1 10" on both fuse and nfs mount
4. add 8 more bricks to the volume with replica count 2

Actual results:
----------------

Fuse mount Output:- mount -t glusterfs 192.168.2.35:/dstore /mnt/gfsc2

[06/04/12 - 15:44:32 root@APP-CLIENT1 gfsc2]# /root/ping_pong_dir/vanilla_ping_pong/ping_pong ./file1 10
unlock at 2 failed! - No such file or directory
lock at 4 failed! - No such file or directory
unlock at 3 failed! - No such file or directory
lock at 5 failed! - No such file or directory
unlock at 4 failed! - No such file or directory
lock at 6 failed! - No such file or directory
unlock at 5 failed! - No such file or directory
lock at 7 failed! - No such file or directory
unlock at 6 failed! - No such file or directory
lock at 8 failed! - No such file or directory
unlock at 7 failed! - No such file or directory
lock at 9 failed! - No such file or directory
unlock at 8 failed! - No such file or directory
lock at 0 failed! - No such file or directory
unlock at 9 failed! - No such file or directory
lock at 1 failed! - No such file or directory
unlock at 0 failed! - No such file or directory
lock at 2 failed! - No such file or directory


Nfs mount output:- mount -t nfs -o vers=3,noac 192.168.2.35:/dstore /mnt/nfsc2

[06/04/12 - 15:44:41 root@APP-CLIENT1 nfsc2]# /root/ping_pong_dir/vanilla_ping_pong/ping_pong ./file1 10
unlock at 1 failed! - No locks available
unlock at 2 failed! - No locks available
unlock at 3 failed! - No locks available
unlock at 4 failed! - No locks available
unlock at 5 failed! - No locks available
unlock at 6 failed! - No locks available
unlock at 7 failed! - No locks available
unlock at 8 failed! - No locks available
unlock at 9 failed! - No locks available
unlock at 0 failed! - No locks available

Expected results:
-------------------
Locks and Unlocks on the region of the file should not fail

Additional info: 
---------------
The file is not self-healed onto new brick. 

Xattrs of file on old brick:-
~~~~~~~~~~~~~~~~~~~~~~~~~~~
[06/04/12 - 16:41:18 root@APP-SERVER1 ~]# ls -l /export_sdc/dir2
total 0
-rw------- 2 root root 11 Jun  4 16:38 file1

[06/04/12 - 16:41:19 root@APP-SERVER1 ~]# getfattr -d -m . -e hex /export_sdc/dir2
getfattr: Removing leading '/' from absolute path names
# file: export_sdc/dir2
trusted.gfid=0x00000000000000000000000000000001
trusted.glusterfs.dht=0x00000001000000009ffffffbbffffff9
trusted.glusterfs.volume-id=0xc95888e0446244bea36c3bd5d8707346

[06/04/12 - 16:41:23 root@APP-SERVER1 ~]# getfattr -d -m . -e hex /export_sdc/dir2/file1 
getfattr: Removing leading '/' from absolute path names
# file: export_sdc/dir2/file1
trusted.gfid=0x497888f6e6f14eeeaa86a77847c176bd

Xattrs on file on new brick:-
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[06/04/12 - 16:36:37 root@APP-SERVER2 ~]# getfattr -d -m . -e hex /export_sdc/dir2
getfattr: Removing leading '/' from absolute path names
# file: export_sdc/dir2
trusted.gfid=0x00000000000000000000000000000001
trusted.glusterfs.dht=0x00000001000000009ffffffbbffffff9
trusted.glusterfs.volume-id=0xc95888e0446244bea36c3bd5d8707346

[06/04/12 - 16:41:35 root@APP-SERVER2 ~]# getfattr -d -m . -e hex /export_sdc/dir2/file1
getfattr: /export_sdc/dir2/file1: No such file or directory

Comment 1 Pranith Kumar K 2012-06-04 10:23:16 UTC
replicate fails lk fop if the errno is not ENOTCONN.This is because it does not have lock healing capability yet.

Comment 2 Niels de Vos 2014-11-27 14:53:39 UTC
The version that this bug has been reported against, does not get any updates from the Gluster Community anymore. Please verify if this report is still valid against a current (3.4, 3.5 or 3.6) release and update the version, or close this bug.

If there has been no update before 9 December 2014, this bug will get automatocally closed.


Note You need to log in before you can comment on or make changes to this bug.