AIX

The Configuration Assistant tries to configure a cluster with Heartbeating over IP interfaces. I.e. the cluster is going to use the non-service addresses AND the Service IP for Heartbeating. For error detection it must be clear from which adapter a heartbeat package has been sent. If there are two adapters in the same network that would not always be clear. Therefore these addresses must be on different subnets which share the same subnet mask. So using

Code:

Node1 -> 192.168.172.124 - 255.255.255.0
Node2 -> 192.168.172.125 - 255.255.255.0
Service-IP 192.168.172.126 - 255.255.255.0

is NOT a valid configuration when using the Config Assistant. If you have to use these addresses you need Hearbeating over IP aliases.

An acceptable configuration for using the Assistant would be

Code:

Node1 -> 192.168.171.124 - 255.255.255.0
Node2 -> 192.168.171.125 - 255.255.255.0
Service-IP 192.168.172.126 - 255.255.255.0

Btw. even some 'old' AIX admins have difficulties when it comes to PowerHA. If you really are a beginner with AIX and your boss expects you to deliver support for the company's most vital server you better ask him to send you on a PowerHA course. This would save you from a lot of headaches.

Thanks for your reply. I've tried your configuration but I'm still getting the same error.

I also tried to use aliases, configuration like this:

Node 1:

Code:

en0: 10.254.245.122
       alias: 192.168.172.124

Node 2:

Code:

en0: 10.254.245.123
       alias: 192.168.172.125

Service IP: 192.168.172.126

Stil getting the same error from the assistant 'no network includes all nodes'.

- Removed existing cluster definition
- Smitty tcpip to assign ip (see /etc/hosts) on both nodes (example node1: IP: 192.168.171.124, subnet: 255.255.255.0)
- /etc/hosts on both nodes:

Code:

127.0.0.1 loopback localhost
192.168.172.126 HTTPlive
192.168.171.124 PHA1 node1
192.168.171.125 PHA2 node2

- Both rhosts empty
- no aliases configured

assistant:

Code:

Communication Path to Takeover Node: PHA2
Application Server: TEST
Application Start Script: /scripts/startHTTP.sh
Application Stop Script: /scripts/stopHTTP.sh
Server IP Label: HTTPlive

Still getting the same error. It's driving me crazy. I've set up a new cluster perfectly without the assistent using this configuration, why won't the assistant do it?

The reason for the Config Assistant was/is that many admins found/find configuring a cluster from the SMIT panels confusing if not difficult. So IBM tried to make this a lot easier by offering to "configure a PowerHA cluster by just answering 5 questions..." However, the price is that the Config Assistant for the basic two-node cluster is limited in both the type of cluster that is being configured and the prereqs it expects. If anything during first node discovery fails (that is where you are at the moment) the Assistant stops.

You probably checked that a hundred times by now but there still is something in the communication between the nodes that causes a problem. Some differences in the serves configuration are difficult to find. Try to let the cluster nodes discover the setup from both sides and look at the messages to find where information is missing/blocked.

Btw what I usually do when installing basic clusters from scratch with the Installation Assistant is that I install a single node plus the PowerHA software. Then I clone this node with mksysb to the secondary node. At that moment I have two nodes which are identical save hostname, uname and addresses.

However, if you can configure a working cluster with the PowerHA SMIT panels you should not need to worry about whether the Assistant works for you or not. In this case you are much more flexible already.