serviceguard refusing communication,cmquercl with alais name

[DeafFrog]

Dear Gurus ,
i am trying this 3 node new cluster.
i have a second NIC of a server set up to have ip 192.168.7.8
made entry to /etc/hosts as (apple is the hostname)
127.0.0.1 localhost loopback
192.168.1.11 apple apple.cd.com
192.168.7.8 orange orange.cd.com
apple:/etc/cmcluster#cmquerycl -C cluster.ascii -n orange -n mango -n banana
Warning: Unable to determine local domain name for apple
Node orange is refusing Serviceguard communication.
Please make sure that the proper security access is configured on node
orange through either file-based access (pre-A.11.16 version) or role-based
access (version A.11.16 or higher) and/or that the host name lookup
on node orange resolves the IP address correctly.
Failed to gather configuration information.
apple:/etc/cmcluster#nslookup 192.168.7.6
Using /etc/hosts on: apple

looking up FILES
Name: banana
Address: 192.168.7.6
Aliases: banana.cd.com

apple:/etc/cmcluster#nslookup 192.168.7.7
Using /etc/hosts on: apple

looking up FILES
Name: mango
Address: 192.168.7.7
Aliases: mango.cd.com

my cmclnodelist list on each of the 3 servsr , that i intend to
install my testclust on is as below:
apple:/etc/cmcluster#more cmclnodelist
orange root
mango root
banana root


i have tested by changing he hostname of apple to orange and cmquerycl goes thru.
Question:
will the cmquerycl accept "hostname" only ?
i tried with ip in cmclnodelist , the error remains the same.

Regards,

[Steven E. Protter]

Shalom,

Make sure the systems are all on the same subnet.

Check the netmask. If you are using standard class C networking, e.g. a netmask is 255.255.255.0, then you have your systems on different networks and that might interfere with the process.

Check /etc/nsswitch.conf that it is using files first for host name resolution.

if changing the name from apple to orange fixes the problem then try this:

Make the hostname of the system in /etc/rc.config.d/netconf match the name in /etc/hosts exactly and that the hosts file is the same on all systems.

SEP

[DeafFrog]

V Salom Steven ,

thanks for the post :


check 1 :
---
apple:/etc/cmcluster#ifconfig lan1
lan1: flags=1843
inet 192.168.7.8 netmask ffffffc0 broadcast 192.168.7.63
apple:/etc/cmcluster#ifconfig lan0
lan0: flags=1843
inet 192.168.1.11 netmask ffffffc0 broadcast 192.168.1.63
apple:/etc/cmcluster#more /etc/nsswitch.conf
hosts:files

---
banana:/etc/cmcluster#ifconfig lan0
lan0: flags=1843
inet 192.168.7.6 netmask ffffffc0 broadcast 192.168.7.63
--
mango:/etc/cmcluster#ifconfig lan0
lan0: flags=1843
inet 192.168.7.7 netmask ffffffc0 broadcast 192.168.7.63

--
check 2
apple:/etc/cmcluster#more /etc/nsswitch.conf
hosts:files
--silarly on other 2--

check 3 ---
apple:/etc/cmcluster#more /etc/hosts |grep -v "^#"
127.0.0.1 localhost loopback
192.168.1.11 apple apple.cd.com
192.168.7.8 orange orange.cd.com
192.168.11.55 bddprotc BDDPROTC BDDPROTC.cd.com
192.168.7.8 bddprotc
192.168.7.6 banana banana.cd.com
192.168.7.7 mango mango.cd.com

mango:/etc/cmcluster#more /etc/hosts |grep -v "^#"
127.0.0.1 localhost loopback
192.168.7.7 mango mango.cd.com
192.168.7.6 banana banana.cd.com
192.168.7.8 orange orange.cd.com

--
banana:/etc/cmcluster#more /etc/hosts |grep -v "^#"
127.0.0.1 localhost loopback
192.168.7.6 banana banana.cd.com
192.168.7.7 mango mango.cd.com
192.168.7.8 orange orange.cd.com

Regards,

[John Bigg · Accepted solution]

If your hostname is apple, then every IP address from apple needs to be listed in the /etc/hosts file on all nodes with a name of apple as an alias. Without this connections will be rejected and you will get the symptoms you see.

You cannot use a different system hostname and network hostname for the same system. It will not work.

[Stephen Doud]

- insure auth/ident and hacl lines are enabled in /etc/inetd.conf and /etc/services and not denied by /var/adm/inetd.sec

- use /usr/newconfig/etc/nsswitch.files as the basis of /etc/nsswitch.conf

- load all fixed IPs hosted by all nodes in the cluster into /etc/hosts and alias each to their sponsoring node's hostname

- restart inetd (inetd -k ; inetd)

- insure /etc/cmcluster/cmclnodelist exists and lists all nodenames and authorizes root access (like a .rhosts file)

- determine if an invalid /etc/cmcluster/cmclconfig file exists on the node and remove it. If the node is presently a member of a cluster, copy the binary file from a functioning node.

- prevent Serviceguard commands from being performed repeatedly in rapid succession by monitoring applications.

[DeafFrog]

Closing the thread , Many thanks to every one for the valuable inputs.
if this thread is googled out :
/etc/host on apple looked like this:

127.0.0.1 localhost loopback
192.168.1.11 apple apple.cd.com
192.168.7.8 orange orange.cd.com
192.168.7.6 banana banana.cd.com
192.168.7.7 cisrnd cisrnd.cd.com

on cisrnd and banana the /etc/entry lokked like this
127.0.0.1 localhost loopback
192.168.7.7 cisrnd cisrnd.cd.com
192.168.7.6 banana banana.cd.com
192.168.7.8 apple orange apple.cd.com orange.cd.com

the cmclnodelist on all the 3 servsr looked like this :
orange root
cisrnd root
banana root


Regards,