cmcheckconf errors

[brian nelson_1]

I am trying to reform a 2 node cluster that was in use some time ago, but hasn't been up in some time. We have had many changes to our network environment since the last time the cluster was up and running. I have made the appropriate changes to the cluster and package ascii files and i am now trying to run "cmcheckconf" to verify these ascii files.
I received the following error messages from "cmcheckconf";
First error is as follows;
Warning: Network interface lan6 on node earth couldn't talk to itself.
Error: No bridged net specified for NM_ID 6 at earth
Second set of errors has to do with the Oracle packages i had previously defined with the cluster. Each node (earth & mars) has one oracle database defined as a package. Since the last time we had the cluster up and running, we have re-IPed and re-subnetted our network. I have change the ascii file for each of the oracle packages to reflect this change, so i'm confused as to where the old 132.10.0.0 subnet is defined. The error messages regarding the packages are as follows;
Error: Invalid subnet 132.10.0.0 in package ORACLE_lear
Error: Invalid subnet 132.10.0.0 in package ORACLE_lear

Our new subnet is 172.16.0.0 and as i said i have changed this in the config files.
Any suggestions on how to remedy these errors?

[Carsten Krege]

1) When an interface cannot talk to itself, you need to check first whether the interface still exists (lanscan). If it does try a linkloop on it.

# linkloop -i

If linkloop doesn't work, SG won't work either. This has then to be fixed.

2) You probably run a "cmcheckconf -C ". Include also the package conf files with

# cmcheckconf -C cluster.ascii -P package1.ascii -P package2.ascii ...

Doing this also the package conf file is updated.

You might want to start from scratch and delete the old cluster binary that still holds the old information that causes the errors:

# cmdeleteconf

Carsten

[Sanjay_6]

Hi Brian,

Sorry abt the earliar posting.

It seems that the error yoou are getting about the network interface lan6 on node earth not being able to talk to itself is because the port is not conneted through a cable to the switch.

If it is connected, the cable might be faulty. This may also happen if you have connected the cable directly to another server through a cross-over cable.

I have the same error when i create the ascii file with the port disconnected from the switch.

Hope this helps

Thanks

[melvyn burnard]

A couple of things to add to th eprevious responses:

1) the subnet you are having a problem with is defined in the pkg config file AND the pkg control file, so you would need to ensure the changes are made to BOTH files, and then the changed files copied to the other node in the cluster.

2) you may want to try doing a cmquerycl -v -C /etc/cmcluster/newconfig.ascii -n earth -n mars
and read the new ascii file to see what changes and configuratiosn ot picks up.

I would then merge any changes you want into the new file, and start with basics, i.e. just do cmcheckconf -v /etc/cmcluster/newconfig.ascii and see whether this errors.

If yes, fix the problems, but if no, then proceed to add in the packages.

HTH

[Sundar_7]

Hi,

Check the stauts lan6 using lanscan. lanadmin and ifconfig.

Run a fresh cmquerycl and create a new config file check running cmcheckconf.

Have you done the changes in Package configuration files and the cluster configuration file ?

Jagadeesh

[Hamdy Al-Sebaey]

hi Brain,
I agree with carsten, just try the following steps mybe it'll solve your problem;
# linkloop -i
# cmcheckconf -C cmclconfig.ascii .
If it does'nt work, then I think it is better to keep a copy of your cmclconf.ascii file & try to issue cmquerycl command to create a new ascii file.
Good luck,
Hamdy