Veritas VolMgr 3.5/Sun Cluster 3.1 scvxinstall problem

• Previous message: kabinet_at_inf.u-szeged.hu: "read-only NFS #2"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

To: <sunmanagers@sunmanagers.org>
Date: Fri, 29 Oct 2004 10:26:04 -0400

Cheers,

I seem to be having a Veritas 3.5 install problem under Sun Cluster 3.1. Has anyone seen this before and have any ideas on what the problem could be? I didn't find anything particularly relevant on SunSolve or the web.

-Thomas

Hardware Platform: Sun Fire 280 R Server
Component Affected: Veritas Volume Manager
OS and Kernel Version: SunOS node1 5.8 Generic_117000-05 sun4u sparc SUNW,Sun-Fire-280R

Describe the problem: [What is wrong with the system?]
Trying to run scvxinstall on one of the nodes (node1) in my Sun Cluster (SC 3.1) and it is failing with the error below. The other node (node2) ran scvxinstall fine. Cluster seems to be fine, when I bring either box down the other knows about it and behaves normally.

node1# scvxinstall

Do you want Volume Manager to encapsulate root [no]? yes

Disabling DMP.
scvxinstall: /dev/vx/dmp is not set up correctly.
scvxinstall: Warning: Unable to disable DMP, but installation will continue...
scvxinstall: /dev/vx/rdmp is not set up correctly.
scvxinstall: Warning: Unable to disable DMP, but installation will continue...
The Volume Manager package installation is already complete.
Cannot validate 315 for vxio. Failed to obtain node information from node2.

[What have you already tried to do?]
I was able to get 1 of the nodes to run scvxinstall successfully (node2). It encapsulated the boot disk and then rebooted several times with no problems. That node is up again running VXVM with no problems. The other node fails when I attempt to run scvxinstall. I checked scstat and everything seems fine to me. I can't see why scvxinstall is failing.

I present two scenarios below that I have tried.

Environment:
Solaris 8
Sun Cluster 3.1
Veritas VM 3.5

As you can see from the error below, scvxinstall is reporting an rpcbind failure connecting to node2 to verify vxio major number. I've tried restarting rpcbind but that doesn't seem to help at all.

Any suggestions?

SCENARIO #1 (no manual entry in /etc/name_to_major):

node1# scvxinstall

Do you want Volume Manager to encapsulate root [no]? yes

Disabling DMP.
The Volume Manager package installation is already complete.
Obtaining the clusterwide vxio number...
scvxinstall: Failed to obtain node information from node2. Trying again...
/usr/cluster/bin/scvxinstall[64]: scrconf: RPC: Rpcbind failure - RPC: Timed out: bad number

SCENARIO #2 (I put a manual entry in /etc/name_to_major to match the entry on the other node (node2) per clustering doco)

node1# scvxinstall

Do you want Volume Manager to encapsulate root [no]? yes

Disabling DMP.
The Volume Manager package installation is already complete.
Cannot validate 315 for vxio. Failed to obtain node information from node2.

Abbreviated output of scstat:

node1# scstat

-- Cluster Nodes --

Node name Status
--------- ------
Cluster node: node1 Online
Cluster node: node2 Online

-- Cluster Transport Paths --

Endpoint Endpoint Status
-------- -------- ------
Transport path: node1:qfe4 node2:qfe4 Path online
Transport path: node1:qfe0 node2:qfe0 Path online

Cluster behavior appears to be normal. I am able to bring down either node of the cluster and I see console messages on the other indicating that everything is switching normally between the nodes.
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers

• Previous message: kabinet_at_inf.u-szeged.hu: "read-only NFS #2"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]