Might be, but before tuning that you should carefully read the TID 10053882
"NetWare Cluster Services: The Gory details of Heartbeats, Split Brains and
Poison Pills"
Klaus
"Alexander Lorenz" <Alexander.Lorenz@4plus.de> wrote in message
news:49005F9B.3C49.0001.0@4plus.de...
> Could it be helpfull to adjust the timeout settings? (Takt, Toleranz,
> Master
> und Slave Watchdog)?
>
> The iSCSI device resides on a HP MSA 1510i connected via a HP Switch.
>
> Thanks so far.
>
> Alex
>
>>>> <arpe@etech.haw-hamburg.de> schrieb am 22.10.2008 um 10:14:
>> Not really as designed. Poison pills happen if one node doesnot answer
>> timely to th heartbeat. As you describe it happens during high load
>> (copy).
>> I would look for communication problems, harddisk driver problems (is
>> the
>> iSCSI device on a seperate box or one of the nodes?).
>> False Poison Pills may also happen due to 2 node Clusters. For 3 and
>> more
>> Nodes it is easier for the nodes to find out which node is alive.
>>
>> Klaus
>>
>> "Alexander Lorenz" <Alexander.Lorenz@4plus.de> wrote in message
>> news:48FE5246.3C49.0001.0@4plus.de...
>>> HI all,
>>>
>>> i have a strange behavior of my two-node cluster (OES2-Linux,
>>> iSCSI-SAN).
>
>>> I
>>> set up a clustered NSS Pool with Volumes. Everything works fine, but if
>> i
>>> trigger a migration from node one to node two during a file copy process
>
>>
>>> on
>>> a Windows XP workstation, the current active node gets a poison pill and
>
>>
>>> is
>>> "killed". The other node is still active and runnig. Is this working as
>>> designed or do i have a problem? If so where should i look for the
>>> reason??
>>>
>>> Thanks in advance
>>>
>>> Alex