MCSG - Changing Subnetmask of Existing System

[Rgomes]

Hi All,

I am giving you the detail. i have 5 servers, 2x rp8400 and 3x rp5470.

I have 2 node cluster running on both rp8400. node1 is my primary and node2 is secandary cell manager. My omni_pkg is running on node1.

Now, I have to change my subnetmask of my existing setup. My present subnet mask
is 255.255.252.0 ( for 10.0.0.0), and this should be 255.255.248.0.

1)What precaution I have to take?
2)And How Can I do it? From SAM?
3)Will it hamper my MCSG configuration? Do I have to halt the cluster or halt specific node before changing the subnetmask of specific server?

Need suggestions/advise from you

Thanks all

Richard

[melvyn burnard]

It depends upon what will happen to your subnet when you make this change. If the subnet chnages, you will have to halt the cluster, make the relevant changes to the networking configuration file(s), and also make any relevant changes to the cluster ascii and package files if the subnet changes as a result of these changes.
You will then need to re-apply the binary and restart the cluster.

If there is no change to the subnet, you will need to halt the cluster, make the changes to the subnet mask, and re-apply the cluster anyway, as this info is kept in the binary and is one of the
cluster parameters you cannot change on-line

[Armin Feller · Accepted solution]

This is the general guidelines for reconfiguring your ip addresses in the Server
and Serviceguard. For details, you should refer to System Admin Manuals for
changing ip address on lancard and MC/Serviceguard manuals for LAN setup in a
cluster environment.

Before making any changes , I would suggest you to make a backup copies of the
following :

Bear in mind that in these examples , the assumption is based on ip addresses
changes ONLY and no other changes involve.

For Servers :

1. /etc/rc.config.d/netconf
2. /sbin/init.d/net
3. /etc/hosts
4. $HOME/.rhosts
5. the output of your "netstat -rn" ( routing tables )

For Serviceguard ( or the entire /etc/cmcluster directory - recommended ) :

5. /etc/cmcluster/cluster.conf (cluster configuration file , it could be other
names at your site)
6. /etc/cmcluster/pkg_name/pkg_name.cntl (package control script)
7. /etc/cmcluster/pkg_name/pkg_name.conf (package configuration file)

These are the necessary actions to perform to make network configurations
changes to HPUX & MC/Serviceguard :

1. Reconfigure the network cards ip addresses

/etc/rc.config.d/netconf

e.g

INTERFACE_NAME[0]=lan0
IP_ADDRESS[0]=15.43.70.148
SUBNET_MASK[0]=255.255.248.0
BROADCAST_ADDRESS[0]=""
INTERFACE_STATE[0]=""
DHCP_ENABLE[0]=0

INTERFACE_NAME[1]=lan1
IP_ADDRESS[1]=1.1.1.1
SUBNET_MASK[1]=
BROADCAST_ADDRESS[1]=""
INTERFACE_STATE[1]=""
DHCP_ENABLE[1]=0

OR you can also use SAM to change the ip address.

2. Reconfigure your routing table

/etc/rc.config.d/netconf

e.g

ROUTE_DESTINATION[0]=default
ROUTE_MASK[0]=""
ROUTE_GATEWAY[0]=15.43.64.1
ROUTE_COUNT[0]=1
ROUTE_ARGS[0]=""

3. Restart your lan ( this could be done by rebooting or manually )

# /sbin/init.d/net start

4. Reconfigure /etc/hosts file

5. Reconfigure $HOME/.rhosts
Ensure that you can rlogin between Serviceguard nodes without having to enter
password.

6. Repeat all the above steps to other nodes in your cluster.

7. Verify your changes :

#lanscan

#ping

#ifconfig lan0
#ifconfig lan1

#netstat -in



Now we will re-configure the Serviceguard configuration files :

1. Make changes to /etc/cmcluster/cluster.conf

e.g

====================================================
NODE_NAME hprc148
NETWORK_INTERFACE lan0
HEARTBEAT_IP 15.43.70.148
NETWORK_INTERFACE lan1
HEARTBEAT_IP 1.1.1.1
FIRST_CLUSTER_LOCK_PV /dev/dsk/c0t1d0
# List of serial device file names
# For example:
# SERIAL_DEVICE_FILE /dev/tty0p0

# Possible standby Network Interfaces for lan0: lan1.

NODE_NAME hprc149
NETWORK_INTERFACE lan0
HEARTBEAT_IP 15.43.70.149
NETWORK_INTERFACE lan1
HEARTBEAT_IP 1.1.1.2
FIRST_CLUSTER_LOCK_PV /dev/dsk/c2t1d0
=====================================================

2. Recreate cluster configuration binary file ( /etc/cmcluster/cmclconfig )

#cmcheckconf -v -C /etc/cmcluster/cluster.conf

#cmapplyconf -v -C /etc/cmcluster/cluster,conf


3. Make changes to /etc/cmcluster/pkg_name/pkg_name.conf

e.g

SUBNET 15.43.64.0


4. Make changes to /etc/cmcluster/pkg_name/pkg_name.cntl

IP[0]="15.43.70.151"
SUBNET[0]="15.43.64.0"

5. ftp or rcp package control file to all nodes which should run this package .


6. Check your configuration :

#cmcheckconf -v -C cluster.conf -P pkg_name.conf


7. If everything is correct , regenerate the cluster config file .

#cmapplyconf -v -C cluster.conf -P pkg_name.conf

8. Start the cluster :

#cmruncl -v

9. Examine the cluster status :

#cmviewcl -v

By now you should see all your nodes up and packages running.

[Rgomes]

Hi Melvyn and Armin,

Thanks both of you. It's very very helpful for me, now I can logically conclude the steps and process.

Armin, the process you mentioned is so detailed, I guess anyone can do it. Thanks.

Melvyn, yes, IP addresses will be in same subnet. So, you have any comment on Armin's steps?


Soon, I am going to apply this on my production servers. So, wish me luck!

Thanks and regards


Richard

[Rgomes]

Hi,

I forgot to assign points for your post.


Regards

Richard