Cmquerycl-problem with A.11.14 for Linux and Redhat 7.3

[Ehr patrick]

I'm trying to set up at two-node cluster and have the following problems:

cmquerycl -C $SGCONF/cluster.ascii -n node1 -n node2 -q quorumnode

complains

a.)about not being able to determine the local domain name

and

b.) Begin checking the nodes...
Looking for other clusters ... Done
Gathering configuration information ............ Done
Failed to probe network
Warning: Network interface bond0 on node node2 couldn't talk to itself.

and

c.) in syslog I find "unable to stat /usr/local/cmcluster/conf/cmclconfig - no such file or directory"

There is a cmclnodelist on every node
I do not use DNS but hosts-files and the interconnect is over a hub.
The nodes can be reached via ping and rlogin from each side over their short-names.

There is only 1 netcard on each node: 3com 905B with bonding bond0-master/eth0-slave

Can anyone point me to the right direction what to check next ??

Thx a lot for every help
Pat

[Uday_S_Ankolekar]

Run cmscancl to troubleshhot this problem. The output of this should give your better picture.

-USA..

[melvyn burnard]

1) check your name resolution, and htat you have set your DNS domain.
2) You have a physical network connectivity problem that must be fixed before you proceed.
Also, you cannot have just one network card, htis is unsupported, you need a second ethernet card to create the bond and have it work.

Be very sure you have correct supported hardware for hte SG version for Linux you ar etrying to run.

[Ehr patrick]

Is network bonding required for ServiceGuard in Linux or is it an option ?

Pat

[Stephen Doud]

SG expects to find the bonding driver... but that doesn't mean you have to set up a standby LAN.

-sd

[Patrick Ehringer]

Hi all,

Problem is solved, cluster is up and running !!

The following changes were made:

to a) added nameresolution with DNS, not only hosts-file and checked nslookup => okay

to b) made again kernel-upgrade to 2.4.18-10 and MANUAL install of serviceguard-rpm files (seems there was problem with sginstall-script)

to c) this error-message is due to the missing deadman-driver!!! The error in syslog is misleading, problem is due to deadman missing!!!
(See HP-DOC Date: 07.08.03
Document description: Error: Unable to establish communication to node [nodename]
Document id: KBRC00012967)

Everything is fine now :-)
Thanks for any help!!
Pat

By the way: hardware for these tests:
2 HP netserver LC3/450
StorageSystem/6
3Com905B-networkcards