Local LAN failover in Serviceguard

[Adam Scheblein]

Greetings,

I am trying to figure out how to setup LAN failover for one of our clusters, we have 4 network cards, so i would like to have a standby for both the heartbeat and the local LAN

I have tried a couple things, however i get the following errors:
# Primary Network Interfaces on Bridged Net 1: lan5.
# Warning: There are no standby network interfaces on bridged net 1.
# Primary Network Interfaces on Bridged Net 2: lan6.
# Warning: There are no standby network interfaces on bridged net 2.

Here is my SG network config:
NODE_NAME Node1
NETWORK_INTERFACE lan5
STATIONARY_IP 192.168.1.101
NETWORK_INTERFACE lan11
NETWORK_INTERFACE lan6
HEARTBEAT_IP 1.1.1.1
NETWORK_INTERFACE lan0

Thanks,
Adam

[Michael Steele_2]

Dear Adam:

Try this link:

http://docs.hp.com/en/B3936-90024/ch05s04.html

[Adam Scheblein]

The example you gave me also has the same warnings that i am getting. Are they OK to ignore?

[Steven E. Protter]

Shalom Adam,

For a card to be standby, it must be configured for the same network as the live card. Most releases of HP-UX won't tolerate two active cards on the same network.

lets look at lanscan
then lanadmin -x for the numeric part of the lan, eg lanadmin -x 5 for lan5

Lets see that we have connectivity and then physically check to see if the physical connection is to the same network.

I think the results will show some wiring or configuration errors.

SEP

[Adam Scheblein]

Also i am getting the following:

Minimum network configuration requirements for the cluster have
not been met. Minimum network configuration requirements are:
- 2 or more heartbeat networks OR
- 1 heartbeat network with local switch (HP-UX Only) OR
- 1 heartbeat network using APA with 2 trunk members (HP-UX Only) OR
- 1 heartbeat network with serial line (HP-UX Only) OR
- 1 heartbeat network using bonding (mode 1) with 2 slaves (Linux Only)


i have read that i could change my stationary_ip to heartbeat and that would be OK, but then why would i have 2 other NIC's for heartbeat?

thanks,
Adam

[Adam Scheblein]

Here is the LAN breakdown:

Lan0 and lan6 are both on the 1.0.0.0 network
lan5 and lan11 are both on the 192.168.1.0 network

I have tested communication on all interfaces, and if i manually set an ip address, i am able to communicate over those links.

What i am trying to do is in serviceguard tell it that the standby for lan5 is lan11

I am no longer worried about lan0 and lan6 standby because i set them both up as redundant Heartbeat cards (one with 1.1.1.1 and the other with 1.1.1.4) which took care of the previous error message i posted.

thanks,
Adam

[Arunkumar.B]

Hi Adam


Try to Configure by this way

NODE_NAME Node1
NETWORK_INTERFACE lan5
HEARTBEAT_IP 1.1.1.1
NETWORK_INTERFACE lan6
HEARTBEAT_IP 192.168.1.101
NETWORK_INTERFACE lan0
NETWORK_INTERFACE lan11

Cheers
Arunkumar.B

[Stephen Doud · Accepted solution]

Standby LAN local switch works between NICs that have the same network stack. For example, all Ethernet cards use the same network stack, so they can be used in any combination for local switching. This means that network interface cards such as 10Base-T, 100Base-T, 100VGAnyLan, 1000Base-T and 1000Base-SX can be combined for local switching.
An example of combinations of cards that cannot be used for local switching are Ethernet cards with FDDI since these cards use different network stacks.
Use 'ioscan -kfC lan' to identify the main drivers involved. iether and igelan are compatible

Serviceguard will not recognize a NIC as a standby if it cannot sustain DLPI traffic (link-layer in OSI model) between it and the primary NIC. Some switches are intelligent enough to be able to block protocols if programmed to do so. Your network specialist should know whether this could be the problem.

Some Core-I/O NICs are not supported with Serviceguard.

VLAN is supported with current versions of Serviceguard (11.14 and up) but perhaps your switch has not been configured to permit the ports to talk to one another.

Have you tried a 'linkloop' command between the primary and expected standby NIC to see if they can communicate?
Example:

# lanscan
Hardware Station Crd Hdw Net-Interface NM MAC HP-DLPI DLPI
Path Address In# State NamePPA ID Type Support Mjr#
0/1/2/0 0x0013215BB502 0 UP lan0 snap0 1 ETHER Yes 119
0/1/2/1 0x0013215BB503 1 UP lan1 snap1 2 ETHER Yes 119
0/3/1/0 0x001185AFB556 2 UP lan2 snap2 3 ETHER Yes 119
0/4/1/0 0x001185AFB579 3 UP lan3 snap3 4 ETHER Yes 119

# linkloop -i 0 0x0013215BB503
Link connectivity to LAN station: 0x0013215BB503
error: get_msg2 getmsg failed, errno = 4
-- FAILED
frames sent : 1
frames received correctly : 0
reads that timed out : 1

# linkloop -i 0 0x001185AFB556
Link connectivity to LAN station: 0x001185AFB556
-- OK

In the latter example, "OK" indicate successful communication.

Note that the "0" in the linkloop command represents the first NIC listed in lanscan (the Crd Instance)

[Bill Costigan]

Your initial setting would be correct if lan5 and lan11 were really on the same network. The message is saying that lan5 is not on a network with any of the other lan cards. [lan6 isn't either]

lan5 and lan11 must be connected to the same ethernet network (i.e. there must be link level connectivity between them) you could put both in the same vlan if you are using vlans.

lan11 and lan0 should not have any IP address configured on them. i.e., they should not show up in a netstat -in

[Adam Scheblein]

Here is the final configuration that worked:

NODE_NAME Node1
NETWORK_INTERFACE lan5
STATIONARY_IP 192.168.1.101
NETWORK_INTERFACE lan11
NETWORK_INTERFACE lan6
HEARTBEAT_IP 1.1.1.1
NETWORK_INTERFACE lan0
HEARTBEAT_IP 1.1.2.1

I needed to specify 2 heartbeat networks and then the remaining NIC was used as a standby for the Data side.

The interesting thing is that i did get an error saying that there were too many NIC's on the original heartbeat network (1.1.1.0)

thanks everyone for the guidance.