unable to configure HPUX MC service guard

[Siva Prasad]

I am trying to setup the cluster in HPUX using MC/serviceguard. I have a common shared disk which is accessed from both the nodes and the public ip address as well as the heartbeat network.

But when i run the command from hpuxdp15(my first node),

cmquerycl -v -C /etc/cmcluster/lpcluster.config -n hpuxdp15 -n hpuxdp14

"unable to establish connection to node hpuxdp15"

Can anyone help me in figuring out my problem.

[Uday_S_Ankolekar]

cmscancl is bestway to start with.. This will show you potential problems with your setup.

Also look for cmclnodelist in /etc/cmcluster has hostname entries for both the nodes.

-USA..

[Geoff Wild]

Run
cmcheckconf -v -C /etc/cmcluster/lpcluster.config

What is output?

Did you follow the "managing MC/ServiceGuard"?

http://docs.hp.com/hpux/onlinedocs/B3936-90073/B3936-90073.html

Rgds...Geoff

[Stephen Doud]

Check out this document:

UMCSGKBRC00008184
TITLE: "unable to establish communications"

(login to the ITRC and access the technical knowledge database - change the search pattern to "Search by document ID"

-sd-

[Victor Fridyev]

Hi,

You have to have root access from computer to another one without password. Put hostnames of the computers to /.rhosts for both of them. I hope, this resolves the problem.

HTH

[Ashwani Kashyap]

create a file called cmclnodelist in /etc/cmcluster and put in it entries like this

first_node root
second_node root
first_node.domain.com root
second_node.doamin.com root

and then try your query again .

You may also want to put both your node names in the .rhosts file as well .

[Sridhar Bhaskarla]

Hi Siva,

As mentioned by Aswini, configure cmclnodelist in /etc/cmcluster directory on both the servers. The syntax is the same that you for .rhosts.

Also, make sure the hostnames listed in the cluster configuration files are in fact the hostnames configured on the servers and are properly resolvable on each server. To verify it enabled inetd with logging

#inetd -l
#tail -f /var/adm/syslog/syslog.log
(in another window)
#telnet $(hostname)

Go the other system and telnet to this system.

Observe the syslog.log to see how the connections are appearing. Do it on both hosts.


-Sri

[Siva Prasad]

I have added all the entries in cmclnodelist but i am still hetting the same error message. I have run the cmscancl command and output is attached. I am able to ping all the heartbeat and public ip from both the nodes.

[Ashwani Kashyap]

Do a linkloop between your interfaces on both the nodes and see if thier is connectivity .

on node1 and node 2 do a lanscan to get MAC address and Crd.inst/PPA number of all LAN cards .

Now from node1 do
#linkloop -v -i crdInt/PPA_of lan0 MAC_address_of_lan0_on_node2

it should come back as ok .

Do it for all lans on bot the nodes .
If thier are failures you needto resolve them first .

[Siva Prasad]

Hi uday,

Can you please check the file i have attached and let me know what is the problem. i have done added all the entries in .rhosts file as well but still getting the same error message. Can you give me your email id.

-siva

[Ashwani Kashyap]

It looks like a hostname resolution problem .

what does hostname command return . It should just retirn the hostname and not the fully qualified domain name .If it does set it just to hostname .

Also check your DNS resolving . Turn it off and do all your resloutin through hosts files unitl the cluster is build .

Also try doing remsh to both the nodes from each other . If asks for passwords then still tehre is a name resolution problems .

[Ashwani Kashyap]

Also check the document that steven mentioned . It has very useful tips to trouble shoot this problem .

[Sridhar Bhaskarla]

Hi,

hpuxdp15.india.hp.com hpuxdp15.india.hp.com
UH 0 lan0 4136
hpuxdp15 hpuxdp15 UH 0 lan1 4136

The above is the problem.

lan1 1500 192.1.1.0 192.1.1.102 124 113
lan0 1500 15.70.177.0 15.70.177.184 19796 16827

Looks like your /etc/hosts is wrong. Both the IPs are pointing to teh same hostname and that is causing the confusion.

-Sri

[Uday_S_Ankolekar]

Try to resolve your hostname problem first.

if you have cmclnodelist configured then .rhsots is ignored. cmclnodelist must have entries of both nodes
nodename1 root
nodename2 root

Does remsh works?
Does netconf file configured full quilified domain name for the host?

-USA..

[Sridhar Bhaskarla]

Hi (Again),

You probably have this configuration in your /etc/hsots file.


15.70.177.184 hpuxdp15.india.hp.com
192.1.1.102 hpuxdp15

Change it to

15.70.177.184 hpuxdp15 hpuxdp15.india.hp.com
192.1.1.102 hpuxdp15-pvt

If you have similar configuration on the other node, change it to ensure uniq names for each interface.

-Sri

PS: Say hi to Dhruva.

[Rita C Workman]

Siva,

I'm 100% with Sridhar on this....looking at your netstat output is just plain 'corn-fusing'.

You need to fix your /etc/host file first....make sure you can reverse lookup everything and get the results (not a conflict) for your host; pkg and heartbeat addresses/names.

Rgrds,
Rita

[Siva Prasad]

My network problems are solved. But i have a different problem now.

I have configured a common shared disk on both the nodes and activated on node one.

when i give the command cmquerycl -v -C /etc/cmcluster/abc.config -n node1 -n node2

it gives me an error message "Unable to determine lvm configuration: failed to receive lvm query reply on node1".

If i give only node2 in the abouve command everything seems to work fine.

Please let me know what can be the reason for this.

[Stephen Doud]

This last problem, "failed to receive lvm query reply " is dealt with in patches for A.11.13 and A.11.14 Serviceguard (you didn't indicate the version you have installed).

A.11.13: PHSS_27722 --> current: PHSS_28849
A.11.14: PHSS_27246 --> current: PHSS_30769
The patches state:

- When this behavior occurs, commands such as:

# cmcheckconf -k -C
# cmapplyconf -k -C

will generate internal errors similar to the following:

Internal error: Number of volume groups sent (20) doesn't
match (1) reply from node hd3grd1
Error: Unable to determine lvm configuration: failed to
receive lvm query reply from node hd3grd1

and then may generate other LVM errors such as:

Error: Volume Group /dev/vgxprdpin is not found on any
nodes in the cluster.
cmcheckconf : Unable to verify cluster file: bad.ascii
========

-sd-