[ofa-general] ib0: multicast join failed

[Yossi Etigin]

On 07/08/09 15:12, Pawel Dziekonski wrote:
> On Fri, 07 Aug 2009 at 03:04:25PM +0300, Yossi Etigin wrote:
>> On 07/08/09 14:25, Pawel Dziekonski wrote:
>>> Hi,
>>>
>>> today I got the following:
>>>
>>> ib0: multicast join failed for ff12:401b:ffff:0000:0000:0000:ffff:ffff, status -11
>>> ib0: multicast join failed for ff12:401b:ffff:0000:0000:0000:ffff:ffff, status -11
>>>
>>> and connection to Lustre was lost.
>>>
>>> I can ping IPoIB address of local iface, but I can't ping any remote IPoIB
>>> address.
>>> There is plenty of free mem so this is not the oom-killer case.
>>> There are no other noticable problems with this host.
>>>
>>> Is it a hardware problem with IB iface?
>> Is your SM alive?
> 
> Well, this is a good question.
> 
> My SM is on Voltaire ISR2012 switch. Today I lost contact with its web
> interface - I don't know why. CLI works fine. Net itself works too. So
> I assume that SM works.
> 

I guess there is some physical problem in the fabric (cables?) because the host
cannot reach the SM - the port is in INIT state.