which interface is used for e.g HA-configuration port 5302 TCP, UDP

[Andre Cornelissen]

In a Metro cluster (R, H and Quorum) the network people need to know which ports to open on which networks.
The Cluster has a Production, Heartbeat and Management network, and the Quorum has Heartbeat and (a different) Management network.
The routes from Production and Heartbeat to Quorum go through a different Firewall than Management.

Questions:
Which NIC is being used for what,
What do we need to the Quorum server.
(it all boils down to: "Which firewall do we need to configure for:")

5300 TCP HA-heartbeat (MC service guard)
5301 TCP HA- general service (MC service guard)
5302 TCP, UDP HA-configuration (MC service guard)
5303 TCP, UDP HA-Probe (MC service guard)
5304 TCP HS-cluster commands (MC service guard)
5305 TCP HA-test (MC service guard)
5408 TCP HA-lock manager (MC service guard)

Regards,
Andre

[Steven E. Protter]

A quorum server deals with which node of he cluster is in charge. If communications and heartbeat are lost a cluster can go into split brain syndrome, where both nodes think they are running the application packages and own the shared disk.

The quorum server decides which node actually gets to do the job. The other node will usually TOC, transfer of control, a controlled crash.

I would run a little report on the NIC cards and look at it to see what NIC does what.

ifconfig lan0
ifconfig lan1

etc

There is also a lot of relavent information in the file /etc/cmcluster/cluster.acii

SEP

[Andre Cornelissen]

Thanks, the config is as follows:

NODE_NAME H
NETWORK_INTERFACE lan2
HEARTBEAT_IP 192.168.129.22
NETWORK_INTERFACE lan8
NETWORK_INTERFACE lan900
HEARTBEAT_IP 192.168.135.70
NETWORK_INTERFACE lan6
#
So apparently the Management lan (lan0) is not used at all, only HB = lan2 (lan8=sb) and Prod. = lan 900 (lan6=sb)

The "Which NIC is used for what" was in a sense of :
"on Heartbeat: 5300 to other node and Quorum,
.....5301 to other node
.....5302 to other node and quorum
on Production: 5304 to other node"
....etc...

Rgds,
Andre

[melvyn burnard · Accepted solution]

Please take a read of this document, it may help.
http://docs.hp.com/en/5874/securingserviceguard.pdf
and SG uses all available NIC's configured in the cluster