[Openais] /etc/init.d/corosync stop hangs

[Steven Dake]

On 03/15/2011 10:53 PM, manish.gupta at ionidea.com wrote:
> Hi Steve,
> 
>        Thanks for your reply. But as suggested in earlier forum threads,
> we need to use iptables to block traffic and make the Ring FAULTY.
> 
>         In this scenario also, we face similar issues.
> 
>         Issue1: When trafiic is blocked for both interfaces simultaneously
> using      iptables -I INPUT -i eth0 DROP
>            iptables -I INPUT -i eth1 DROP
> 

Ya blocking traffic on both interfaces will probably result in some kind
of unwanted behavior.  I recommend at this time to use bonding for
deployments.  If you really want to use redundant ring, we need some
others from the community to contribute to that development - I can't do
it alone.

>         Finally if the ethernet for interface or Ring(eth0 or eth1) is
> ready to allow traffic, then also Corosync will not detect that
> Ring to be UP.
> 

Automatic ring recovery is a work item that is not yet finished.  There
is a patch floating around which needs some more work.  Lars had
indicated he would find someone to wrap up the patch, or perhaps you can
if you feel the urge :)

>         Issue2: Multiple Heartbeat processes starts, thus makng the system
> very slow when corosync-cftool -r command is issued when both
> Interfaces are down. Or if 2 interfaces go down, and the Interface
> which went down Firstly (using iptables)is made UP..
> 
>          Please help......
> 

Redundant ring is not supported in upstream at this time.  If you can
write code, supply patches.  If not, then using bonding.  The community
will sort out these problems eventually but it is a big job and needs
more contribution from people.

Thanks
-steve

> 
> On Tue, March 15, 2011 8:29 pm, Steven Dake wrote:
>> ifconfig and redundant ring don't work well together.
>>
>> In fact, you should avoid ifconfig down altogether.
>>
>>
>> http://www.corosync.org/doku.php?id=faq:ifdown
>>
>>
>> Regards
>> -steve
>>
>>
>> On Tue, Mar 15, 2011 at 6:32 AM, <manish.gupta at ionidea.com> wrote:
>>
>>
>>> Hi,
>>> I have configured two node cluster with 2 channels
>>> Corosync stop command is not working and system is hanged every time
>>> Configuration and logs details are mentioned below for my scenario.
>>>
>>>
>>>
>>> corosync.conf file. content
>>>
>>> totem { version: 2
>>> secauth: off
>>> threads: 0
>>> rrp_mode: passive
>>> interface { ringnumber: 0
>>> bindnetaddr: 192.168.4.0
>>> mcastaddr: 225.0.0.1
>>> mcastport: 694
>>> }
>>> interface { ringnumber: 1
>>> bindnetaddr: 192.168.5.0
>>> mcastaddr: 225.0.0.2
>>> mcastport: 694
>>> }
>>> }
>>>
>>>
>>> logging { to_syslog: no
>>> fileline: off
>>> to_stderr: no
>>> to_logfile: yes
>>> to_syslog: yes
>>> logfile: /var/log/corosync.log
>>> debug: on
>>> timestamp: on
>>> }
>>>
>>>
>>> amf { mode: disabled
>>> }
>>>
>>>
>>> #end corosync.conf
>>>
>>>
>>> Network card
>>> etho 192.168.4.46 eth1 192.168.5.46
>>>
>>>
>>>
>>> After ifconfig eth0 down
>>> ifconfig eth1 down operation . Corosync is continuously writting these
>>> log in every three second interval. 1. corosync [TOTEM  ] Token ws lost
>>> in commit state 2. corosync [TOTEM  ] ntering GATHER state from 4
>>> 3. corosync [TOTEM  ] craeting commit token because I am is the rep
>>> 4. corosync [TOTEM  ] storing new sequence id for ring 6cb4#[used to be
>>> changed] 5. corosync [TOTEM  ] enetring commit state
>>>
>>>
>>>
>>>
>>> After making down both channel command
>>> /etc/init.d/corosync stop is not working
>>>
>>>
>>> No new log is written  for corosync stop
>>>
>>>
>>> After executeing corosync stop command
>>> ps -afe |grep coro command show these process
>>>
>>> corosync [corosync]<defunct>
>>> grep coro
>>>
>>> Version Details
>>> corosync-1.3.0
>>>
>>>
>>> If I configure only one channel then everything is woking fine.
>>> Please can you let me know my testing scenarion is valid or not.
>>>
>>>
>>> _______________________________________________
>>> Openais mailing list
>>> Openais at lists.linux-foundation.org
>>> https://lists.linux-foundation.org/mailman/listinfo/openais
>>>
>>>
>>
> 
> 
> _______________________________________________
> Openais mailing list
> Openais at lists.linux-foundation.org
> https://lists.linux-foundation.org/mailman/listinfo/openais