On 12/13/2010 02:18 PM, Florian Haas wrote:
> On 12/13/2010 08:15 PM, Steven Dake wrote:
>> On 12/13/2010 12:08 AM, Florian Haas wrote:
>>> On 12/07/2010 01:13 PM, Florian Haas wrote:
>>>> Honza, Steve,
>>>>
>>>> do I understand correctly that this would trigger if, for example, a
>>>> user had configured an iptables rule that blocked UDP port 5405? And if
>>>> yes, would Corosync behave any differently when instead there is no
>>>> _physical_ link between nodes on cluster startup?
>>>>
>>
>> Yes you can close your bug.
>>
>> This only effects the case where corosync can't even communicate with
>> itself on the local node.  This usually occurs because iptables are
>> enabled incorrectly.  It could also possibly happen when a nic fails.
>> In either of those cases, corosync can now be terminated without a kill -9.
> 
> OK, so do I understand correctly that
> 
> * "iptables -A INPUT -p udp --dport 5405" would trigger the issue,
> 

yes

> whereas
> 
> * disabling both node's ports on the switch would not?
> 

I am not certain on this second point, i am on the road atm, i'll give
it a try and let you know tuesda.

Regards
-steve

> Cheers,
> Florian
> 
> 
> 
> 
> _______________________________________________
> Openais mailing list
> [email protected]
> https://lists.linux-foundation.org/mailman/listinfo/openais

_______________________________________________
Openais mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/openais

Reply via email to