How to change the ip addresses for both nodes(Service Guard) in a clustered environment

[Mike Vizza]

I need to re-ip both HP-UX servers which are using MC Service Guard. Is there a document I can reference?

[Jeff Schussele]

Hi Mike,

Well, not that I know of, But what you need to do is stop the pkgs & then the cluster & run:

set_parms ip_address

& then make sure *all* hosts files as well as DNS records are updated. IF you're moving *out* of the org subnet, you'll need to change any & all virtual IPs as well. That command will require a reboot.

HTH,
Jeff

[Ermin Borovac]

You will probably need to update cluster configuration file with new IPs (and package config file if you are changing any of the relocatable IPs).

Have a look in the following technical knowledge base document.

Document description: How to change a ServiceGuard cluster's HEARTBEAT IP
Document id: UMCSGKBRC00012626

[Stephen Doud]

There is a document for this:
http://ITRC.HP.COM document ID: UMCSGKBRC00009727
TITLE: Changing LAN information in a ServiceGuard cluster


The cluster binary file (/etc/cmcluster/cmclconfig) gives Serviceguard
the profile of your cluster when the cluster forms.

Using cmviewconf, you can inspect the content of the cluster binary file
and notice the IPs of the NICs are loaded in the binary file.

Some of the IP-related information comes from the cluster configuration
ASCII file and some from the package configuration ASCII files located
in /etc/cmcluster and lower.

To update the cluster binary with the new IP information:
1) halt the cluster (cmhaltcl -f)
2) disable auto-cluster-formation at boot time: /etc/rc.config.d/cmcluster: AUTOSTART_CMCLD=0 (set on each node)
3) edit /etc/rc.config.d/netconf on each node - updating the NIC's IP/netmask info
4) update the hostname resolution service files with the new IP
5) verify nslookup matches the hostname of each node to the new IP
6) use ifconfig to replace the IP - reboot if needed
7) update the cluster and package configuration files with the new IP/Subnet as needed
8) update the package control scripts with new relocatable IPs if necessary
9) cmapplyconf -f -C
If this fails, use 'cmdeleteconf -f' to zero the cluster binary, then repeat the above command
10) cmapplyconf -f -P -P ...
11) use cmviewcl to verify the cluster binary is up to date
12) cmruncl (start the cluster - test package startup
13) re-enable cluster-auto-start:
/etc/rc.config.d/cmcluster: AUTOSTART_CMCLD=1

-StephenD.