cmquerycl : Unable to initialize interfaces error (Linux version)

[Tal Moran]

Hello,

I just install serviceguard on Linux Redhat ES 3.0 according to the book (B9903-90033). When I try to build the cluster configuration file I get this error:

# cmquerycl -v -L /dev/cciss/c0d1p1 -n lab-sth-1-cls-1 lab-sth-1-cls-2 -C $SGCONF/lpcluster.config
cmquerycl : Unable to initialize interfaces, No such file or directory

When I start the serviceguard I get this message:

# /etc/init.d/cmcluster.init start
tcp 0 0 0.0.0.0:5302 0.0.0.0:* LISTEN
udp 0 0 0.0.0.0:5302 0.0.0.0:*
cmrunnode : Unable to initialize interfaces, No such file or directory
ERROR: Unable to join cluster

System information:
2* HP DL380 G3 Connected to MS500 SCSI base system.
Red Hat Enterprise Linux ES release 3 Update 1, kernel version 2.4.21-9.
Serviceguard RPM install in the system:
-serviceguard-A.11.15.02-99.product.redhat.i386.rpm
-sgcmom-B.02.02.01-99.product.redhat.i386.rpm

Thanks,
Tal.

[melvyn burnard]

wel I dont know if this is a typo, but you have a -n option missing in your line, it should read:

-n lab-sth-1-cls-1 -n lab-sth-1-cls-2 -C $SGCONF/lpcluster.config

Also, why are you trying to run serviceguard if you have not created a cluster by applying the binary yet?
And doing /etc/init.d/cmcluster.init start is not the way tostart it, you should do
cmruncl

[Tal Moran]

Hi,

According to the documentation (Chapter 5 page 157)
http://www.docs.hp.com/linux/pdf/B9903-90033.pdf

first I need to build the cluster configuration file using cmquerycl command.
after that I can build the binary configuration using ( page 164):
cmapplyconf -v -C $SGCONF/lpcluster.config

When I run any command I get this error:
# cmruncl
cmruncl : Unable to initialize interfaces, No such file or directory
# cmviewcl
cmviewcl : Unable to initialize interfaces, No such file or directory

Thanks,
Tal.

[Serviceguard for Linux]

Have you confirmed that you can rlogin from system to system?

Do you have bonding set up? If so, which NIC drivers are you using? If lsmod shows tg3, then that may be your problem. The release notes indicate that the BCM5700 must be used.

Another common mistake is that cmclnodelist (and rhosts) are not correctly set up. See the section Editing Security Files (chapter 5) in the Users (Managing) manual

[Tal Moran]

Hi,

I can rlogin from system to system.
I configure the cmclnodelist file and .rhosts file according to the documentation(Chapter 5).
The nslookup work OK from the two computers.
I use the BCM5700 driver for the NIC.
And still I get this error.

Thanks Tal.

[Serviceguard for Linux]

So I assume that the partition for the Quorum disk is the correct type (83 if I remember correctly).

One other useful thing to try to help isolate the problem is to do a cmquerycl for the LockLun by itself and then the nodes by themselves. As I remember, these are legal ways to use cmquerycl. Obviously, when trying the nodes individaully, run the command on that node.

Another thing, it shouldn't matter but 2.4.21-9.0.1 or 2.4.21-9.0.3 are the actual supported errata. If you do update, remember to rebuild the deadman driver.

If you still have the problem, you probably want to attach the log files.

[Tal Moran]

Hi
I found the problem it was my mistake, I add a line to /etc/cmcluster.conf. I add this file to the bash_profile so I add export line in this file.
Thanks for youâ re help
Tal