cmcheckconf fails on new cluster config - please help

[ConnieK]

HPUX 11.23 ServiceGuard v. A.11.18.

I am new to SG, so please forgive any ignorant questions!

I am having issues getting cmcheckconf to complete successfully so I can begin this cluster. It is failing on the network part of the config file. Here is the message:

"Non-uniform connections detected,
server2 lan7 10.10.1.2 successfully received from server1 lan7 10.10.1.1
but server1 lan7 10.10.1.1 did not receive from server2 lan7 10.10.1.2.
This could be due to heavy network traffic, or heavy load on server2.
Failed to evaluate network
cmapplyconf: Unable to reconcile configuration file /etc/cmcluster/newcluster.config
with discovered configuration information."

I have established the nodelist, so I do not use .rhosts files. There is no traffic on server2 as these are brand new servers - still in the build process.

Linkloop from server2 to server1:
linkloop -i 7 0x001A4BF38311
Link connectivity to LAN station: 001A4BF38311
-- OK

Linkloop from server1 to server2:
linkloop -i 7 0x001A4BF382BD
Link connectivity to LAN station: 001A4BF382BD
-- OK

server1:
ifconfig lan7
lan7: flags=1843
inet 10.10.1.1 netmask ff000000 broadcast 10.255.255.255


server2:
ifconfig lan7
lan7: flags=1853
inet 10.10.1.2 --> 255.255.255.0 netmask ff000000 broadcast 10.255.255.255


Please help me figure this out. Thanks in advance!

[Steven E. Protter]

Shalom,

Check ping and other network connectivity on these networks.

server2 lan7 10.10.1.2 successfully received from server1 lan7 10.10.1.1
but server1 lan7 10.10.1.1 did not receive from server2 lan7 10.10.1.2.

Normally the cause of this is simple. Loose network cable, switch configuration blocks traffic, stuff like this.

SEP

[Bill Costigan]

Why is the configuration different on the two severs? server2 has the POINTTOPOINT option and server1 doesn't

What does the entries in /etc/rc.config.d/netconf look like.

BTW server1's config looks more correct than server2's.

Bill

[ConnieK]

Stephen - I cannot ping the other HB IP from either server.

From server2 to server1:
ping 10.10.1.1
PING 10.10.1.1: 64 byte packets
45 packets transmitted, 0 packets received, 100% packet loss


From server1 to server2:
ping 10.10.1.2
PING 10.10.1.2: 64 byte packets
3 packets transmitted, 0 packets received, 100% packet loss


I checked the cables. No loose ones.

Maybe switch problems.....hmmm....

Bill - I saw that output from ifconfig and I have NO idea how it got that flag "pointtopoint" in there.

Output of netconf:

server1:
IP_ADDRESS[7]=10.10.1.1
SUBNET_MASK[7]=255.0.0.0
INTERFACE_NAME[7]=lan7
BROADCAST_ADDRESS[7]=10.255.255.255
INTERFACE_STATE[7]=up

server2:
IP_ADDRESS[7]=10.10.1.2
SUBNET_MASK[7]=255.0.0.0
INTERFACE_NAME[7]=lan7
BROADCAST_ADDRESS[7]=10.255.255.255
INTERFACE_STATE[7]=up

Connie

[ConnieK]

I just checked with the network person - the vLan that the Heartbeat sits in is not routed. He checked the ports from his side and both are active and connected - no port errors.

Could it be the flag "pointtopoint"?

[Steven E. Protter]

Shalom again:
Stephen - I cannot ping the other HB IP from either server.

It's Steven BTW.

Fix the ping connectivity issue and then try to run the configuration script again.

SEP

[ConnieK]

If the vLan these two IP's sit in is not routed, then how can I ping them successfully?

[Bill Costigan]

Sounds like you may have other 10.0.0.0 NICs configured.

perhaps the ping the doesn't work is trying to use another NIC.

The subnet mask is strange for a heartbeat nic. I would think 255.255.255.0 or even 255.255.255.248 might be more approprate.

please send a netstat -in from both servers.

[ConnieK]

Bill - I can change the masking on these cards, no problem, but I don't think that has too much to do with this issue. I ran an arp command on both servers and it looks to me that one of the cards thinks it should route - see below - but I don't think it should:

Server2 # arp -a n
server1 (###.#.##.###) at 0:1a:4b:f3:96:59 ether
core1-gw (###.#.##.###) at 0:0:c:7:ac:0 ether
10.10.1.2 (10.10.1.2) -- no entry


Server2# netstat -in
Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll
lan3 1500 ###.#.##.0 ###.#.##.## 3132 0 24 0 0
lo0 4136 127.0.0.0 127.0.0.1 61933 0 61933 0 0
lan7 1500 10.0.0.0 10.10.1.2 229 0 171 0 0
lan900 1500 ###.#.##.0 ###.#.##.### 819358 0 139950 0 0



Server1 # arp -a n

server1 (###.#.##.###) at 0:1a:4b:f3:97:a9 ether
core1-gw (###.#.##.###) at 0:0:c:7:ac:0 ether


Server1:# netstat -in
Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll
lan3 1500 ###.#.##.0 ###.#.##.## 3129 0 46 0 0
lo0 4136 127.0.0.0 127.0.0.1 34501 0 34501 0 0
lan7 1500 10.0.0.0 10.10.1.1 44 0 102 0 0
lan900 1500 ###.#.##.0 ###.#.##.### 845859 0 159364 0 0

[Bill Costigan]

How about the routing tables on both servers.

netstat -rn

[Bill Costigan]

I'm also assuming that lan3 and lan900 are not in the 10.x.x.x network.

You have those IP addresses blocked out so I cannot tell.

Basically you want to ensure the all traffic to the 10.x.x.x (based on the netmask) goes out the lan7 nic and directly to the other server without trying to go though a router or gateway.

[ConnieK]

Bill,

Correct, lan3 and lan900 are not in this network. The "10" netwokr is just a "made-up" Ip that sits in it's own vLan - not routed, just like a cross-over cable. This is a heartbeat network set up specifically for this purpose.

On server1 - netstat -rn (only for lan7):

10.10.1.1 10.10.1.1 UH 0 lan7 4136

10.0.0.0 10.10.1.1 U 2 lan7 1500


ON server2 - netstat -rn (only for lan7):
10.10.1.2 10.10.1.2 UH 0 lan7 4136

255.255.255.0 10.10.1.2 UH 2 lan7 1500


Does it look like that server2 Ip is trying to route itself?

[Bill Costigan]

You must have some bogus entry in the netconf file on server2 or maunally entered some ifconf command.

on server2 there is no route to the 10.0.0.0 network. just a route to two hosts:
10.10.1.2 and 255.255.255.0

On server1 - netstat -rn (only for lan7):
10.10.1.1 10.10.1.1 UH 0 lan7 4136
10.0.0.0 10.10.1.1 U 2 lan7 1500


ON server2 - netstat -rn (only for lan7):
10.10.1.2 10.10.1.2 UH 0 lan7 4136
255.255.255.0 10.10.1.2 UH 2 lan7 1500

Perhaps there is some goofy routing entries in netconf.

let's see all the routing entries in netconf. also I don't know why there is a route metric of '2' for the route to the 10.0.0.0 network. it really should be a '1'

[Bill Costigan]

did you enter your own 'route add' command?

If netconf looks good you could re-execute it by typing
/sbin/init.d/net start

[Bill Costigan · Accepted solution]

By editing the posts you may be hiding the problem.

For the netstat -in output you say for lan7 only.

There may in fact be a route to 10.0.0.0 pointing to lan3 which you removed form the posting and which says to send all the 10.0.0.0 traffic out lan3.

[ConnieK]

Bill,

Well, you were correct! There wasn't a bogus entry in the netconf file, but I remember that I did do an "ifconfig" last week and probably messed it up. I finally decided to reboot the server to clean up the mess and look , see!!!!

cmcheckconf: Verification completed with no errors found.
Use the cmapplyconf command to apply the configuration.

YYEEEHAW!

Now onto even more fun with making the packages!

Thanks so much!

[Bill Costigan]

Great. Glad to hear it's working.

[ConnieK]

No - there was no routing to Lan3. I needed to edit (sorry) for confidentiality.