cmquerycl: unable to obtain information for: .

[LUIS F. GARCIA JARA]

Hello everybody !

I have a 2-node cluster (HP9000 K460 and D370) running HP-UX 10.20 and MC/ServiceGuard A.10.06, with PHSS_21994 (patch for MC/ServiceGuard)

The cluster is up and running with no problem, however, when I try to run a cmquerycl with no arguments from any of the nodes, I get the following message:
-----
All nodes within the cluster will be checked.
cmquerycl : Unable to obtain information for: .
Connection timed out.
----
I've checked the lan connections for both nodes and they're ok. (Each node has two lan cards and they reach each other with no problem using linkloop, within the node and out).

I've also checked name resolution for both nodes using nslookup and they're responding fine. My .rhosts file uses the correct node names for both nodes also.

Is there any other place I can look for a possible bad configuration ?
Anybody has a clue ?

I'd appreciate your help.

Thanks.
Luis Fernando.

[linuxfan]

Hi Luis,

This may not be related to your problem directly, but talks about a similar problem on a D-class server. The document ID is :G2616190

/Begin/
Problem Description

When performing the command:

cmquerycl -v -C /etc/cmcluster/cmclconf.ascii -n node1 -n node2

Begin checking the nodes...
All nodes within the cluster will be checked.
I receive the error message:

.cmquerycl : Unable to obtain information for: .
Connection timed out.

What is the problem?

Configuration Info

Operating System - HP-UX
Version -10.20
Hardware System - HP 9000
Series -D370

Solution

The problem here is lan0 has a hardware state UP and software
statis DOWN and no lan cable is applied between the nodes. So
lan0 has no cable connection. The solution is to apply the
crossed Twisted pair cable:

Pinout: 1 ---> 3 \_ one twisted cable pair
2 ---> 6 /
3 ---> 1 \_ second twisted cable pair
6 ---> 2 /

or use a hub.

/End/

In your situation, i am not sure which lan cards you are using.

-HTH
Ramesh

[LUIS F. GARCIA JARA]

Hi, Ramesh !

Thanks for your response, but the cluster is using its own hub, which is isolated from other servers, and the communication betweeen nodes is working fine. I had read the document you mention, but unfortunately, that's not my case.

Luis Fernando

[Sundar_7]

Hi,

cmquerycl mostly depends on Name Resolution.

U might be using DNS ..

So check out /etc/resolv.conf entires.

Try using nslookup and see if it resolves the node names..

This might help U out..

See if it helps and assign the points :-)

Sundar

[Sundar_7]

Oh ! sorry

I did nt notice that U already checked out
using nslookup..

Check Out /etc/cmcluster/cmclnodeslist

Also I believe it uses some port in /etc/services..

Check if they have been enabled

Sundar

[linuxfan]

Hi,

You said the cluster is already up and running.
Does cmquerycl fail or cmquerycl -n node1 -n node2 fail also?

what about cmscancl?

Do you have any lan cards with hardware state Up and software state down?

-Ramesh

[melvyn burnard]

Sorry to correct the previous responder but 10.06 of ServiceGuard does not have cmscancl, nor does it use /etc/cmcluster/cmclnodelist.

I would suggest you confirm that the networking configuration is correct, including such things as broadcast addresses etc.
Ensure that nslookup can resolve the host names BOTH ways, i.e. lookup ip as well as lookup hostname.
I would also suggest you try one each node the following:
cmquerycl -v -n node_name_of_node_you_are_logged_on_to

and see if the problem is related to one node or both.

[LUIS F. GARCIA JARA]

Hi,

First of all, thank you all for your help.

Ramesh:
You're right, the cluster has no problem starting/stopping, but I can no longer reconfigure it, as cmapplyconf uses cmquerycl, and it fails alone or with any parameters.
I have two lan cards and one of them is up/up and the other is up/down, being the latter the stand-by interface.

Melvin:
I checked the net config and is ok, IP addresses, netmasks, etc.
nslookup works fine in both nodes using hostname and IP address, and cmquerycl does not work on either node.

I'm starting to think there might be a problem with some of the patches I lastly installed, and it might not be one related to MC/SG...

-Luis

[Mladen Despic]

Luis, just a thought:

Have you tried running 'remsh ' from and vice versa? This should verify that your .rhosts are configured the way you intended them to be.

[Mladen Despic]

I guess you could also try 'rcp' since that's what MC/SG tries to do.

[LUIS F. GARCIA JARA]

Mladen:
I checked remsh and rcp and they work fine both ways.

I've just restored a make_recovery tape I did before my last patch upgrade, and without those patches cmquerycl works ok. The patch list is:

PHKL_22855, PHKL_23612, PHKL_23836, PHKL_23838, PHKL_23962, PHNE_22507, PHNE_22877, PHCO_22268, PHCO_23042, PHCO_23437, PHCO_23684, PHCO_23844, PHCO_23861, PHNE_17103, PHNE_23948, PHSS_22354, PHSS_22652, PHSS_22944, PHSS_23535, PHSS_23798, PHSS_23804

One (or some, maybe) of them is screwing up things.

I'm going to reinstall them one by one to find out which one is.

Thanks.
-Luis