Unable to establish comm... : 19.

[Evan_25]

I've read and read each thread about having trouble with cmquerycl, and still have not resolved the issue.

I'm using SLES8, SG A.11.15.
here is the output of the initial cmquerycl:
Looking for other clusters ... Done
Error: Unable to establish communication to node ppi2: 19
Failed to gather configuration information.

Here is the contents of my cmclnodelist:
ppi1 root
ppi2 root

I am using DNS, and I can "host" each node and get the correct ip back. I can rlogin/ssh etc.. to each node from each node, and bypass the password prompt.
I am using 2 nic cards in each DL560G2. One card is directly connected to the other node via a peer-to-peer cable. The other nic's connect the LAN. I am not using the channel
bonding driver, is this required?

The only other message I can get out of cmquerycl is from /var/log/messages:
Sep 19 14:08:20 PPI2 cmclconfd[23268]: Unable to stat /opt/cmcluster/conf/cmclconfig, No such file or directory.

Can anyone give me anymore hints,suggestions, or pointers? I would greatly appreciate it!

Thanks,
-Evan

[Evan_25]

Well, after strace'ing cmquerycl, I got a hint to match the case of the nodes to match their names in /etc/hosts.

All seems well now. That sure was a frustrating day!!

[Srinivas Krishnappa Sha · Accepted solution]

Hello Evan,

There are a couple of things that you may have to verify in order to get this problem solved.

Which version of serviceguard patch level are you on ? I know that we released a patch for 11.15 , 11.15.06 in the last week.
So you may need to update it.

1) Verify that the deadman module is loaded.
lsmod | grep deadman

2) Verify that the pidentd module is loaded. lsmod | grep pidentd

3)Verify that the process identd is started.
ps -ef | grep iden

4) Channel bonding driver is not required.

5) Make sure that each of the nodes have the cmclnodelist file populated or they have the etc/hosts and .rhosts entries.

6) Reboot the systems and verify that the modules are still loaded. (deadman and pidentd)

7) Verify that the process identd is started.
ps -ef | grep iden

8) If you find that the modules are not loaded after restrat ( use the comaand )
insserv MODULENAME
to make sure that the module is loaded at that init level every time that system is restarted.

9) Now try the cmquerycl

10) If all of these does not work, try updating the patch level (11.15.06) latest.

11) Remove the Back to Back cable between the 2 nodes and then try to do the same thing.

12) Still not working, Please uninstall and reinstall OS and Serviceguard.

Do let me know if this solves your problem.

[njia_1]

Hi

I had this problems before. It is because I did a typing error when I configured the cmclconf.ascii file. It took me a while to find out.

Check all the hostnames and Ip address in cmclconf.ascii file.

[Evan_25]

It actually turned out that the case of the host names matter when using cmquerycl.

In the hosts file they names where in upper case, whereas in the cmquerycl command I was using lower case in the host names.

Thanks for everyones help!!

[njia_1]

Hi Evan,

Assigning points to all the answers is a better way to say thanks

[Evan_25]

Thanks for everyones help!