Re: Split Brains - Not another Boy Band

Rajesh,

you should get your ha logs (on aix /var/ha/logs), but most important for instance evictions are the lmd and lms trace files as they list the actual reason that OWS can decode correctly.

Remember that a simple thing like time synchronization can be a cause for instance eviction, it happened to us earlier this week.

interconnect issues is another.
Raj

On Thu, 3 Mar 2005 17:10:29 -0500, Rajesh.Rao_at_jpmchase.com <Rajesh.Rao_at_jpmchase.com> wrote:
> Hello All, Not being much of an SA, are they any specific OS or Cluster
> logfiles that I can look for, to determine the cause of one node being
> evicted in a two node cluster (9.2.0.5.0 RAC on Solaris 5.8 and Sun Cluster
> 3.0), due to a split brain. The eviction happened due to "reason 3", which
> would indicate a communication failure, across the interconnect. Nothing in
> /var/adm/messages or syslog. We suspect there was heavy resource
> utilization on one of the nodes, which could have caused it.
>
> Thanks
> Raj
>
> --
> http://www.freelists.org/webpage/oracle-l
>

-- 
------------------------------
select standard_disclaimer from company_requirements where category =
'MANDATORY';
--
http://www.freelists.org/webpage/oracle-l