HACMP problem

[Guest_imported]

In an rotating configuration cluster i get a problem.
when take over activity take place the service address of the node
which is taking over resources shows 0.0.0.0 .......however all
other interfaces looks as they are expected.
Once tcpip deamons are refreshed problem get solved.
Any ideas???????????????

 

[prv3116]

First, where are you seeing the 0.0.0.0? In a netstat output or what? Is the fall-over successful, resources/applications/etc moving over to the other node?

There is so much to HACMP, particularly regarding the interfaces configurations, both regular AIX configs need to be checked and the HACMP configs need to be checked. Please list just a bit more detail and I'll see if I can give you some ideas.

 

[Guest_imported]

hi
Infact i see 0.0.0.0 on service adapter which should have on take over
address after take over.
The takeover appears to be good except the problem with that
single problem.
regds
Triu

 

[prv3116]

What version of HACMP? Did you install it or are you assuming the ownership of a cluster already configured? How many nodes in the cluser? Does the problem occur only when going from nodeA to nodeB, or does it occur going either direction?

Check your error log to see if you are getting any specific adapter errors logged.

Check your ODM configuration for the adapters on all nodes involved in the cluster. Use "smitty chginet" and confirm (and make note of) the IP address and Network Mask and Current State (make sure Current State is UP). Use "smitty mktcpip" and confirm (and make note of) everything here (hostname, IP, Mask, gateway, nameserver). DO THIS ON ALL NODES IN THE CLUSTER.

Now check all these configurations in your /etc/hosts, /.rhosts. Hostnames and IP addresses must be exact matches on all nodes. Both short and long names should be in the .rhosts file. DO THIS ON ALL NODES IN THE CLUSTER.

Verify hostnames and IP addresses with your DNS nslookup and also ping each by IP and by hostname.

Check your /etc/resolv.conf and /etc/netsvc.conf

Now check these same settings in your HACMP configurations. Check your setting in Cluster Topology for the Network Modules, Network Adapters.

Check your /tmp/hacmp.out log after a fall-over and see what is happening during the IPAT (IP Address Takeover) for the service adapter.

Run a cluser Verification to see if there are any errors logged. Use "smitty hacmp" then go to Cluster Configuration, then Cluster Verification. Run the Verify Topology and then the Resources separatly.

Try running the HACMP synchronization, particulary if it only occurs in a fall-over from nodeA to nodeB.

Check out the online documentation at:

http://www.rs6000.ibm.com/doc_link/en_US/a_doc_lib/aixgen/hacmp_index.html

If you are on older than 4.3.1, you'll have to located hardcopy.

Hope this helps. I know it is a lot to do, and I am listing these as if you had little HACMP knowledge, so please pardon me if you are more advanced. This is a good system checkup to do periodically anyway, even on so-called "stable" clusters. It helps you to ensure that nothing has changed and confirms your knowledge of your systems.


Remember, amateurs built the Ark, professionals built the Titanic!