[SunHELP] Sun Cluster questions

Nadine Miller vraptor at promessage.com
Tue May 18 10:01:52 CDT 2004


Darryl Brooks wrote:
> Nadine -
> 
> Did your /etc/path_to_inst file change?
> 
Nope.  I checked that after bringing it up into multi-
user again with the -x flag.

We ended up calling up Sun, and they recommended bringing
node1 up in single-user with the clustering services still
on.  This worked (I hadn't tried it fearing the hang that
I got in multi-user with clustering on).

In the process of trouble-shooting, something we did (in
notes at office--an scconf command, iirc) seemed to tickle
node0, and caused it to crash.  The support guy recommended 
immediately that I ^d the node I was working on to bring
it into multi-user mode (thinking, I guess that they would
both have to re-negotiate).

Both came up cluster aware.  They were still a little con-
fused, so we ended up doing two scswitch commands to
push the node master back and forth between the two
machines.  They seem copacetic now.

Unfortunately, /var on node0 did not have enough space to
get a crash dump. :-/  But I have transferred explorer
output.  We'll see if that's enough to get root cause.

Both boxes need patching and the T3s they are connected
to need FW updates.  If only the frigging clients would
give me the breathing space to do my real job--
infrastructure maintenance--instead of forcing me and
my team to QA their crappy vendor-built software. :-(

It's too bad we didn't have real downtime--at least
that would have put the fear of the computer ghods
into them.

=Nadine=

> 
> ---- Original message ----
> 
>>Date: Mon, 17 May 2004 12:29:51 -0700
>>From: "N. Miller" <vraptor at promessage.com>  
>>Subject: [SunHELP] Sun Cluster questions  
>>To: sunhelp at sunhelp.org
>>
>>Brief background:
>>
>>Had to take 1 node of a two node SC3.0 cluster down to
>>replace an I/O board.  The node1 was removed using scswitch
>>before shutting down.  The cluster failed over to node0 as 
>>expected.  
>>
>>Node1's hardware was repaired, and we attempted to bring it 
>>up to connect to the cluster, but instead it just sat and 
>>spun it's wheels at:
>>
[snip]



More information about the SunHELP mailing list