topic Re: Node is refusing Serviceguard communication in Operating System - Linux

Node is refusing Serviceguard communication

George Denyer_1 — Tue, 12 Feb 2008 17:36:16 GMT

Hello,

I am yet another victim of this error message: "Node XYZ is refusing Serviceguard communication".

I am setting up a cluster consisting of two nodes running RedHat EL 5 and ServiceGuard A.11.18. The cluster hasnt been created yet, I'm trying to run cmquerycl for the first time and I get this message:

[root@cafsa1 conf]# cmquerycl -v -C $SGCONF/cafsa.conf -n cafsa1
Warning: Unable to determine local domain name for cafsa1
Looking for other clusters ... Done
Node cafsa1 is refusing Serviceguard communication.
Please make sure that the proper security access is configured on node
cafsa1 through either file-based access (pre-A.11.16 version) or role-based
access (version A.11.16 or higher) and/or that the host name lookup
on node cafsa1 resolves the IP address correctly.
Failed to gather configuration information.
[root@cafsa1 conf]#

After reading the docs and the many posts on this forum related to similar problems I have made sure the firewall is disabled, identd is running, cmclnodelist lists the node with user root (I also tried with "+"), the hosts file lists the node correctly, and just in case /root/.rhosts also lists the node with user root; and yet, I keep getting the same error message.

I seem to have ran out of ideas, so if anybody has something to suggest, I'll appreciate it.

Re: Node is refusing Serviceguard communication

John Bigg — Wed, 13 Feb 2008 08:53:04 GMT

Make sure that all IP addresses on the system resolve to the hostname. SG should allow aliases also but the 11.18 base code does not. You will need a patch for that.

Check that xinetd is running. Is it even installed? rpm -q xinetd I believe that RH5 may not install it by default and if it's not running you will get the symptom you see. Can you telnet localhost on port 5302? i.e.

telnet localhost 5302

If not this indicates an xinetd/SG cmclconfd problem and although it does not test the UDP connection which looks like it could be broken this will at least show you if things are fundamentally broken.

Future rpms will have a dependency on the xinetd rpm.

If all this checks out then post your syslog file.

Re: Node is refusing Serviceguard communication

George Denyer_1 — Wed, 13 Feb 2008 17:08:01 GMT

Hello John, thanks for your answer.

All IP addresses on the system do resolve to the hostname, without aliases; but I'm pretty sure xinetd is not installed, if it's a requirement that's probably it.

I'll install it today and test. Thanks.

Re: Node is refusing Serviceguard communication

Steven E. Protter — Wed, 13 Feb 2008 18:09:00 GMT

Shalom,

Serviceguard uses xinetd to run a couple of daemons. If its been commented out or not installed, this will need to be corrected before moving forward.

The install should have squawked about this however.

SEP

Re: Node is refusing Serviceguard communication

George Denyer_1 — Thu, 14 Feb 2008 00:05:47 GMT

Hello all,

I installed xinetd, as John suggested, and the error is gone.

Apparently xinetd is not installed by default in RHEL5, SG doesnt verify if it is installled, and the docs dont mention it as a requirement.

Thanks to John and Steven.

G. Denyer