[Ocfs2-users] SuSe Hangs when /etc/init.d/o2cb online

[Jan Kara]

On Fri 04-04-08 17:29:11, Miguel Santos N. wrote:
> Well, my problem is partially solved.
>
> I upgraded the kernel on the Opensuse nodes the cluster comes online 
> smoothly =)
>
> Howeeeeever....
>
> When I try to mount the Filesystem, this message comes up:
>
> =====
> drwagner:~ # mount -t ocfs2 /dev/sdf1 /mnt/projects/
> mount.ocfs2: Transport endpoint is not connected while mounting /dev/sdf1 
> on /mnt/projects/. Check 'dmesg' for more information on this error.
> [....]
> (4600,2):o2net_check_handshake:1149 node arenamexico (num 0) at 
> 192.168.130.1:7777 advertised net protocol version 103 but 8 is required, 
> disconnecting
> (9301,2):dlm_request_join:920 ERROR: status = -107
> (9301,2):dlm_try_to_join_domain:1069 ERROR: status = -107
> (9301,2):dlm_join_domain:1342 ERROR: status = -107
> (9301,2):dlm_register_domain:1533 ERROR: status = -107
> (9301,2):ocfs2_dlm_init:2345 ERROR: status = -107
> (9301,2):ocfs2_mount_volume:1087 ERROR: status = -107
> ocfs2: Unmounting device (8,81) on (node 2)
> =====
  I think this is the problem I warned about: OCFS2 has changed the network
protocol incompatibly so either all your nodes have to run old kernel or
all nodes have to run a new kernel. So you either have to upgrade your
CentOS kernel to 2.6.22 or newer, or you have to try running OpenSUSE with
and older kernel...

> At the first sight I thought "hmm... maybe it is the version of the 
> ocfs2-tools", however I downgraded the ocfs2-tools on the CentOS server, 
> and I remounted, and the problem still...
>
> If I mount the OCFS2 on a Opensuse server, the problem exists on the Centos 
> server, tha makes me think in the possibility of some sort of 
> misconfiguration when I formated the FS, but I have no idea.
>
> Any hint?

								Honza
> Jan Kara escribió:
>>> Hello,
>>>
>>> I have a DELL MD3000i, and a couple of servers that I want to connect to
>>> the array.
>>>
>>> I have setup the main server with CentOS 5.
>>> - 2.6.18-53.el5 x86_64
>>> - ocfs2-tools-1.2.7-1.el5
>>> - ocfs2console-1.2.7-1.el5
>>> - ocfs2-2.6.18-53.el5-1.2.8-2.el5 # Kernel Module
>>>
>>> And two OpenSuSe 10.3 Servers
>>> - 2.6.22.5-31-default x86_64
>>> - ocfs2-tools-1.2.6-18
>>> - ocfs2console-1.2.6-18
>>> The module has been already shipped with the kernel (thanks to novell)
>>>     
>>   Hmm, as far as I remember I also had some problems with older OpenSUSE
>> 10.3 kernels and OCFS2 (resulting in hangs). Can you try with the latest
>> kernel from OpenSUSE 10.3?
>>   Also I'm actually not sure whether 2.6.18 from RH and 2.6.22 from
>> OpenSUSE will play well together - there were some incompatible network
>> protocol changes but I'm not completely sure when exactly did they take
>> place and what RH took into their kernel.
>>
>> 								Honza
>>
>>   
>
-- 
Jan Kara <jack at suse.cz>
SUSE Labs, CR