[Gluster-users] [gluster] possible split-brain issue

Fri Jan 23 08:48:37 UTC 2015

Hi Pranith,

No worries!

Here is the output of the other brick:

[root at dmf-wpst-2 ~]# getfattr -d -m. -e hex /export/vdb1/brick/
getfattr: Removing leading '/' from absolute path names
# file: export/vdb1/brick/
security.selinux=0x756e636f6e66696e65645f753a6f626a6563745f723a66696c655f743a733000
trusted.afr.gv0-client-0=0x000000000000001500000000
trusted.afr.gv0-client-1=0x000000000000000000000000
trusted.gfid=0x00000000000000000000000000000001
trusted.glusterfs.dht=0x000000010000000000000000ffffffff
trusted.glusterfs.volume-id=0x51de44c3f01e486da6b710c7b7a270d7

[root at dmf-wpst-2 ~]#  getfattr -d -m. -e hex /export/vdb1/brick/mpdis/
getfattr: Removing leading '/' from absolute path names
# file: export/vdb1/brick/mpdis/
security.selinux=0x756e636f6e66696e65645f753a6f626a6563745f723a66696c655f743a733000
trusted.afr.gv0-client-0=0x000000000000000200000000
trusted.afr.gv0-client-1=0x000000000000000000000000
trusted.gfid=0x8ff7afeb996244cd9d1bf213568398d7
trusted.glusterfs.dht=0x000000010000000000000000ffffffff

[root at dmf-wpst-2 ~]# getfattr -d -m. -e hex /export/vdb1/brick/mpdis/test.rep.00.00.00.00.dmf1
getfattr: Removing leading '/' from absolute path names
# file: export/vdb1/brick/mpdis/test.rep.00.00.00.00.dmf1
security.selinux=0x756e636f6e66696e65645f753a6f626a6563745f723a66696c655f743a733000
trusted.afr.gv0-client-0=0x000000000000000000000000
trusted.afr.gv0-client-1=0x000000000000000000000000
trusted.gfid=0x85ed306b179b46819d7c02eb336543b8

[root at dmf-wpst-2 ~]# getfattr -d -m. -e hex /export/vdb1/brick/mpdis/test.rep.00.00.00.00.dmf2
getfattr: Removing leading '/' from absolute path names
# file: export/vdb1/brick/mpdis/test.rep.00.00.00.00.dmf2
security.selinux=0x756e636f6e66696e65645f753a6f626a6563745f723a66696c655f743a733000
trusted.afr.gv0-client-0=0x000000000000000000000000
trusted.afr.gv0-client-1=0x000000000000000000000000
trusted.gfid=0xa826a389e7a042c2b5175a1acbecae9b

From: Pranith Kumar Karampuri [mailto:pkarampu at redhat.com]
Sent: Friday, January 23, 2015 3:42 PM
To: Arnold Yang; Jifeng Li; Gluster-users at gluster.org
Subject: Re: [Gluster-users] [gluster] possible split-brain issue

hi Arnold,
   You gave the output only on one brick it seems? Could you also provide it on other brick as well. Sorry I didn't make that clear in my earlier mail.

Pranith
On 01/23/2015 10:44 AM, Arnold Yang wrote:
Hi Pranith,

Here is the output for the commands provide by you, anything more you need, please tell us!
Thanks!

[root at dmf-wpst-1 ~]# getfattr -d -m. -e hex /export/vdb1/brick/
getfattr: Removing leading '/' from absolute path names
# file: export/vdb1/brick/
security.selinux=0x756e636f6e66696e65645f753a6f626a6563745f723a66696c655f743a733000
trusted.afr.gv0-client-0=0x000000000000000000000000
trusted.afr.gv0-client-1=0x000000000000001400000000
trusted.gfid=0x00000000000000000000000000000001
trusted.glusterfs.dht=0x000000010000000000000000ffffffff
trusted.glusterfs.volume-id=0x51de44c3f01e486da6b710c7b7a270d7

[root at dmf-wpst-1 ~]# getfattr -d -m. -e hex /export/vdb1/brick/mpdis/
getfattr: Removing leading '/' from absolute path names
# file: export/vdb1/brick/mpdis/
security.selinux=0x756e636f6e66696e65645f753a6f626a6563745f723a66696c655f743a733000
trusted.afr.gv0-client-0=0x000000000000000000000000
trusted.afr.gv0-client-1=0x000000000000000400000000
trusted.gfid=0x8ff7afeb996244cd9d1bf213568398d7
trusted.glusterfs.dht=0x000000010000000000000000ffffffff

[root at dmf-wpst-1 ~]# getfattr -d -m. -e hex /export/vdb1/brick/mpdis/test.rep.00.00.00.00.dmf1
getfattr: Removing leading '/' from absolute path names
# file: export/vdb1/brick/mpdis/test.rep.00.00.00.00.dmf1
security.selinux=0x756e636f6e66696e65645f753a6f626a6563745f723a66696c655f743a733000
trusted.afr.gv0-client-0=0x000000000000000000000000
trusted.afr.gv0-client-1=0x000000000000000000000000
trusted.gfid=0x85ed306b179b46819d7c02eb336543b8

[root at dmf-wpst-1 ~]# getfattr -d -m. -e hex /export/vdb1/brick/mpdis/test.rep.00.00.00.00.dmf2
getfattr: Removing leading '/' from absolute path names
# file: export/vdb1/brick/mpdis/test.rep.00.00.00.00.dmf2
security.selinux=0x756e636f6e66696e65645f753a6f626a6563745f723a66696c655f743a733000
trusted.afr.gv0-client-0=0x000000000000000000000000
trusted.afr.gv0-client-1=0x000000000000000000000000
trusted.gfid=0xa826a389e7a042c2b5175a1acbecae9b

From: Pranith Kumar Karampuri [mailto:pkarampu at redhat.com]
Sent: Thursday, January 22, 2015 12:14 AM
To: Jifeng Li; Gluster-users at gluster.org<mailto:Gluster-users at gluster.org>; Arnold Yang
Subject: Re: [Gluster-users] [gluster] possible split-brain issue

On 01/14/2015 04:48 PM, Jifeng Li wrote:
Hi ,

[issue]:  To ensure the glusterFS mount point work, a script will periodically using HTTP put a file to subdirectory under mount point which is used as  Apache DocumentRoot. But after running some time,  some errors show below:
[2015-01-14 09:18:40.915639] E [afr-self-heal-common.c:2868:afr_log_self_heal_completion_status] 0-gv0-replicate-0:  metadata self heal  failed,   on /mpdis
[2015-01-14 09:18:41.924584] E [afr-self-heal-common.c:233:afr_sh_print_split_brain_log] 0-gv0-replicate-0: Unable to self-heal contents of '/' (possible split-brain). Please delete the file from all but the preferred subvolume.- Pending matrix:  [ [ 0 20 ] [ 21 0 ] ]
[2015-01-14 09:18:41.925182] E [afr-self-heal-common.c:2868:afr_log_self_heal_completion_status] 0-gv0-replicate-0:  metadata self heal  failed,   on /
[2015-01-14 09:18:41.934827] E [afr-self-heal-common.c:233:afr_sh_print_split_brain_log] 0-gv0-replicate-0: Unable to self-heal contents of '/mpdis' (possible split-brain). Please delete the file from all but the preferred subvolume.- Pending matrix:  [ [ 0 4 ] [ 2 0 ] ]
[2015-01-14 09:18:41.935375] E [afr-self-heal-common.c:2868:afr_log_self_heal_completion_status] 0-gv0-replicate-0:  metadata self heal  failed,   on /mpdis
[2015-01-14 09:18:42.943742] E [afr-self-heal-common.c:233:afr_sh_print_split_brain_log] 0-gv0-replicate-0: Unable to self-heal contents of '/' (possible split-brain). Please delete the file from all but the preferred subvolume.- Pending matrix:  [ [ 0 20 ] [ 21 0 ] ]
[2015-01-14 09:18:42.944432] E [afr-self-heal-common.c:2868:afr_log_self_heal_completion_status] 0-gv0-replicate-0:  metadata self heal  failed,   on /
[2015-01-14 09:18:42.946664] E [afr-self-heal-common.c:233:afr_sh_print_split_brain_log] 0-gv0-replicate-0: Unable to self-heal contents of '/mpdis' (possible split-brain). Please delete the file from all but the preferred subvolume.- Pending matrix:  [ [ 0 4 ] [ 2 0 ] ]
[2015-01-14 09:18:42.947323] E [afr-self-heal-common.c:2868:afr_log_self_heal_completion_status] 0-gv0-replicate-0:  metadata self heal  failed,   on /mpdis
[2015-01-14 09:18:43.955929] E [afr-self-heal-common.c:233:afr_sh_print_split_brain_log] 0-gv0-replicate-0: Unable to self-heal contents of '/' (possible split-brain). Please delete the file from all but the preferred subvolume.- Pending matrix:  [ [ 0 20 ] [ 21 0 ] ]
[2015-01-14 09:18:43.956701] E [afr-self-heal-common.c:2868:afr_log_self_heal_completion_status] 0-gv0-replicate-0:  metadata self heal  failed,   on /
[2015-01-14 09:18:43.958874] E [afr-self-heal-common.c:233:afr_sh_print_split_brain_log] 0-gv0-replicate-0: Unable to self-heal contents of '/mpdis' (possible split-brain). Please delete the file from all but the preferred subvolume.- Pending matrix:  [ [ 0 4 ] [ 2 0 ] ]

Besides, I find  Input/output error shown below  when listing the files of under mount point:

[root at dmf-wpst-2 mpdis]# ll

total 0

-rwxr-xr-x. 1 apache apache 0 Jan 14 04:21 test.rep.00.00.00.00.dmf1

-rw-r--r--. 1 apache apache 0 Jan 14 04:21 test.rep.00.00.00.00.dmf2

[root at dmf-wpst-2 mpdis]# ll

total 0

-rwxr-xr-x. 1 apache apache 0 Jan 14 04:21 test.rep.00.00.00.00.dmf1

-rw-r--r--. 1 apache apache 0 Jan 14 04:21 test.rep.00.00.00.00.dmf2

[root at dmf-wpst-2 mpdis]# ll

ls: cannot open directory .: Input/output error

[root at dmf-wpst-2 mpdis]# ll

ls: cannot access test.rep.00.00.00.00.dmf1: Input/output error

ls: cannot access test.rep.00.00.00.00.dmf2: Input/output error

total 0

?????????? ? ? ? ?            ? test.rep.00.00.00.00.dmf1

?????????? ? ? ? ?            ? test.rep.00.00.00.00.dmf2
    Any tips about debugging further or getting this fixed up would be appreciated.

[version]: 3.5.3
[environment]:  two virtual server each has one brick :

root at dmf-wpst-2 mpdis]# gluster volume status

Status of volume: gv0

Gluster process                                                                                                Port       Online   Pid

------------------------------------------------------------------------------

Brick dmf-ha-1-glusterfs:/export/vdb1/brick                      49152    Y              332

Brick dmf-ha-2-glusterfs:/export/vdb1/brick                      49154    Y              19396

Self-heal Daemon on localhost                                                  N/A        Y              19410

Self-heal Daemon on 10.175.123.246                                       N/A        Y              999

[root at dmf-wpst-1 mpdis]# gluster volume info

Volume Name: gv0

Type: Replicate

Volume ID: 51de44c3-f01e-486d-a6b7-10c7b7a270d7

Status: Started

Number of Bricks: 1 x 2 = 2

Transport-type: tcp

Bricks:

Brick1: dmf-ha-1-glusterfs:/export/vdb1/brick

Brick2: dmf-ha-2-glusterfs:/export/vdb1/brick

Options Reconfigured:

nfs.disable: ON

network.ping-timeout: 2

storage.bd-aio: on

storage.linux-aio: on

cluster.eager-lock: on

performance.client-io-threads: on

performance.cache-refresh-timeout: 60

performance.io-thread-count: 64

performance.cache-size: 8GB

cluster.server-quorum-type: none

  [mount-point info]:

1.       mount command

glusterfs -p /var/run/glusterfs.pid --volfile-server=dmf-ha-1-glusterfs --volfile-server=dmf-ha-2-glusterfs --volfile-id=gv0 /dmfcontents

2.       mount point directory hierarchy
[root at dmf-wpst-2 /]# ls -ld /dmfcontents/
drwxr-xr-x. 5 root root 71 Jan 14 04:39 /dmfcontents/
               [root at dmf-wpst-2 /]# ls -ld /dmfcontents/mpdis/
              drwxr-xr-x. 2 apache apache 89 Jan 14 04:39 /dmfcontents/mpdis/
hi Jifeng Li,
     Sorry for the delay in response. Could you post the output of:
'getfattr -d -m. -e hex <brick-path>'
'getfattr -d -m. -e hex <brick-path>/mpdis'
'getfattr -d -m. -e hex <brick-path>/mpdis/test.rep.00.00.00.00.dmf1'
'getfattr -d -m. -e hex <brick-path>/mpdis/test.rep.00.00.00.00.dmf2'

Pranith

_______________________________________________

Gluster-users mailing list

Gluster-users at gluster.org<mailto:Gluster-users at gluster.org>

http://www.gluster.org/mailman/listinfo/gluster-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150123/40a111c1/attachment.html>