Changing the heartbeat IP

[Anthony_90]

I have a 2 node cluster. I have a plan to change the IP address. So I modified the cluster conf file. Before changing the IP address of heartbeat LAN and before bringing down the service guard I want to make sure the changes I made is not going to make the cmapplyconf to fail. So I ran the

cmapplyconf -C

and got the following message,

Begin cluster verification...

Error: Network interface lan0 on node starr1 has a different IP address (11.10.1
0.1 != 10.10.10.3)
Error: Network interface lan0 on node starr2 has a different IP address (11.10.1
0.2 != 10.10.10.4)
Error: lan0 on node starr1 cannot be configured in the cluster
because it does not have an IP address, and it is not a standby lan for any oth
er lan.
Error: lan0 on node starr2 cannot be configured in the cluster
because it does not have an IP address, and it is not a standby lan for any oth
er lan.
Failed to evaluate network
cmcheckconf : Unable to reconcile configuration file clust_conf.ascii.new
with discovered configuration information. I understand the first two error messages because the heartbeat LAN has not been setup with the new IP address.

Why I'm getting the last two error messages?..

Error: lan0 on node starr1 cannot be configured in the cluster
because it does not have an IP address, and it is not a standby lan for any oth
er lan.
Error: lan0 on node starr2 cannot be configured in the cluster
because it does not have an IP address, and it is not a standby lan for any oth
er lan.


Thanks
Anth Biji

[Anthony_90]

I wanted to add one more question. Do I have to run "cmapplyconf -P " if I modify the package control scripts.

[Chris Watkins_1]

You will definitely have to bring down the cluster first.
Since the nic cards are already in use by the running binary,
ServiceGuard is saving you from yourself, so to speak.
If you're talking about the IP address of the node itself,
I wouldn't even consider making the changes with ANY software up
and running, personally. So many things "can" break that way.
(I feel less strongly about "package" IP's, but you were unclear)
While you're cmhalt'd... that's a good time to change the package scripts.
You "can" change the package control scripts with the cluster up
and running... but you'd be better off not to, unless you're
positive you've thought it through very carefully.

Halt the packages, halt the cluster.
Do the cmcheckconf to see if you have issues with the cluster config.
If not, make your package changes, do your cmapplyconf, and
start the cluster up again. You should be good to go from there.

[Sridhar Bhaskarla]

Hi,

1. 'cmapplyconf' is going to fail when the cluster is running because the heartbeat IPs you specified in your cluster ascii files are not associated with the interfaces yet. So, bring down the cluster, change the IPs on the interfaces and then apply the configuration. Before applying the configuration, make sure you get a response from all the nodes when you ping to the heart-beat subnet.

2. You don't have apply serviceguard configuration if you changed the cluster control file. Only changes to cluster ascii or the package ascii (conf) file will require you to re-apply the configuration.

-Sri

[Bernhard Mueller]

Anthony,

actually you may want to download
http://docs.hp.com/hpux/onlinedocs/B3936-90065/B3936-90065.html

attached is a list of cluster/package changes and required cl/pgk state (extract from the doc).

BTW, I would not worry too much about the wording of the error message (as long as it points you in the right direction).

Regards,
Bernhard

[Stephen Doud]

When working with Serviceguard (or troubleshooting any HPUX subsystem or kernel problem - always pay attention to the very first error message. The first error message is usually the reason for the succeeding error messages. You cannot eliminate the later error messages until after correcting the first error. I call this the cascading error effect.

In this case, the first error indicates a difference in IP addresses for the NIC:

Error: Network interface lan0 on node starr1 has a different IP address (11.10.1
0.1 != 10.10.10.3)

This is expected since the NIC does not have the new IP assigned to it! You will not get that to change as long as NIC has the old IP assigned to it.

So:
1) first update the /etc/rc.config.d/netconf file with the new IP.

2) ifconfig the new IP onto the NIC

3) halt the cluster

4) cmapplyconf the updated ASCII file.

-> If cmapplyconf still fails on this issue, use "cmdeleteconf -f" to zero out the cluster binary on all nodes and decluster the shared VGs

-> cmapplyconf the updated cluster ASCII, along with the package ASCII file(s).

4) cmruncl

-sd

[Anthony_90]

Thanks for your response. But I'm looking for the answers on my last two error messages,

Error: lan0 on node starr1 cannot be configured in the cluster
because it does not have an IP address, and it is not a standby lan for any oth
er lan.
Error: lan0 on node starr2 cannot be configured in the cluster
because it does not have an IP address, and it is not a standby lan for any oth
er lan.

lan0 and lan1 has the IP address. Why it's reporting it doesn't have the IP address?

[Chris Watkins_1]

What you're trying to do will not work.
*That* is why you're getting the error.

As has been mentioned already, you sometimes need to look at
FIRST errors before worrying about others that follow.
That's where the problem lies.
You already have lan0 configured.
cmapplyconf sees this, and is preventing further action.