Thanks to everyone that responded to the inquiry.
The Solution:
"Just as you recommended the 2 memory channel cards need to be on the
same bus... I took one of the cards out and with no changes the system cam
up."
There were also a lot of other good suggestions that I will list here:
(Thanks to Bryan Williams, Timothy Brown, Paul LaMadeleine and Manish Vashi)
1) Please check the jumper setting on memory channel for window size it
should be
128 for ES45's ,as if you have dual memory channel controller in one
system it
does not support the window size of 512.This is because on ES45 both
the memory
channel adapters sit on the same PCI bus and if you have window size
of 512 it
exceeds the bandwith.We had the same problem with a 6 node cluster 3
ES40 and 3
ES45,where the ES45 used to crash after adding the member.
go this link
http://www.tru64unix.compaq.com/docs/updates/ES45/CHCNFGRN.HTM#sec-128mb
2) I found this using google and searching for " cfs_mountroot_local "
http://aa11.cjb.net/tru64_unix_managers/2000/10/0290.html
3) I've seen this before. Check to make sure the quorum disk is visible
on the
storage controller. I also would check that your connections on the
controller are set to Tru64, not Windows NT.
4) You know about the restriction about 2 MC cards in ES45's?
(Just curious) What version of the HSG firmware are you using? (ACS
is most current version )
We ran into a similar problem with V5.1 on ES40's a couple of weeks
ago.
Once the problem occurred, however, nothing worked. Attempting to
boot
into the cluster after that would fail at the point of trying to
access
the quorum disk.
What cleared the condition was to reboot the HSG controller using
the
button on the controller. We saw which one was serving the quorum
and
cluster disks and rebooted it, causing the disks to fail over to the
other controller. Suddenly everything worked. Later we failed the
disks
back over and everything stayed normal...
The Original Problem was:
>Has anyone ever seen this before?
>
> My customer is installing TruCluster V5.1A with (2) ES45 systems and
>FC storage. The customer has redundant MC adapters (already been checked)
>with redundant MC Hubs.
> The customer successfully installs TruCluster, and can add the 2nd
>cluster member.
> Then when booting the 2nd member, both systems lock up when the 2nd
>system hits the quorum.
> Each Member has 1 vote and the quorum has 1 vote.
> See additional information below supplied by the customer.
>
>Thanks,
>
>Matt Morris
>Technical Support Engineer
>Avnet TECenter
>7300 W. Detroit Street
>Chandler, AZ 85226-2410
>
>Matt,
>
>And yes I'm booting the second system from the generic kernel, and yes I
>have a shared storage subsystem (hsg80) that the quorum and cluster
>filesystems reside on.
>
>Node 1 comes up fine, I can log in cluster looks ok, I boot node 2 kernel
>loads services are loading find till I get here
>
>BEGIN
>
>registering CMS Service
>ics_mct: icsinfo set for node 2
>ics_mct: Declaring this node 2
>CNX QDISK: adding 1 quorum disk vote toward formation
>
>#### OTHER SYSTEM LOCKS UP HERE
>
>CNX MGR: Cluster cluster_name incarnation 0xe0b30 has been formed
>CNX MGR: founding node is 2 csid is 0x10001
>CNX MGR: membership configuration index: 1( 1 addition, 0 removed)
>CNX MGR: node NODE2 2 incarn 0xe0b30 csid 0x10001 has been added to cluster
>dlm: resuming lock activity
>kch: resuming activity
>CNX QDISK: successfully claimed quorum disk, adding 1 vote
>CNX MGR: quorum (re)gained, (re)started cluster operations
>joining versw kch set
>clsm: checking peer configurations
>clsm: initialized
>waiting for cluster mount to complete
>warning: cfs_perform_glroot_mount: cfs_mountroot_local failed to mount the
>cluster root fs with error = 6
>
>
>END
>
>Thanks again,
>Jason Hedden
>
Received on Thu Dec 20 2001 - 22:08:07 NZDT