cluster config. problem

[manish_csc]

Hi Friends !!

I am facing an issue as per my cmgetconf file config on one node shown as below :-

NODE_NAME XXXXX
NETWORK_INTERFACE lan0
HEARTBEAT_IP xxx.xxx.xxx.xxx
NETWORK_INTERFACE lan1
HEARTBEAT_IP xxx.xxx.xxx.xxx
NETWORK_INTERFACE lan2
NETWORK_INTERFACE lan3

As per above config is that ok lan2 config. as standby for lan0.

Lan0 and Lan2 briged on same network and checked conectivity ok by linkloop shown ok.

Is that makes any difference to define standby interface in last instead of just after next line of Lan0.


Thanks in Advance !!

Manish

[skt_skt]

cmgetconf output depends on how it is defined on cluster.ascii file. If it is NOT live bring down the primary interface using UNIX command and watch syslog.log; it will tell where is it getting failed over; enble the interface and it will fail back too.

[manish_csc]

This is live production , 2 days back i faced the problem when Lan0 failed , this was not switched to standby interface. All packages failover to alternate node.

[Uday_S_Ankolekar]

I would put just below the Primary Network or may be just after Hearbeat.
NODE_NAME
NETWORK_INTERFACE lan0
HEARTBEAT_IP
NETWORK_INTERFACE lan2 <--- added

-Uday

[manish_csc]

My concern is only that if is there any difference if

NODE_NAME XXXXX
NETWORK_INTERFACE lan0
HEARTBEAT_IP xxx.xxx.xxx.xxx
NETWORK_INTERFACE lan1
HEARTBEAT_IP xxx.xxx.xxx.xxx
NETWORK_INTERFACE lan2
NETWORK_INTERFACE lan3

OR

NODE_NAME XXXXX
NETWORK_INTERFACE lan0
HEARTBEAT_IP xxx.xxx.xxx.xxx
NETWORK_INTERFACE lan2
NETWORK_INTERFACE lan1
HEARTBEAT_IP xxx.xxx.xxx.xxx
NETWORK_INTERFACE lan3

As per standard config file should defined as last one but my config file having entries as first 1.
If lan0 failed what happened in both cases ???????????

[skt_skt · Accepted solution]

i observed both type of entries on my cluster.ascii file and both works fine.

My observation is all the standby interface are on the same Bridged net ID as the primary one.

Since it is new implementation i would ask to cross check the all the Bridged net ID for all inerfaces.Also check with the network team at the switch side/physical connection..

Also cross check with the link loop.Also format and read /var/adm/nettl.LOG000/0001 files if available.

are the lan cards dual/multi port capable?Shift to a different port on the same card.

[Stephen Doud]

When you use cmquerycl to create a cluster configuration template file, Serviceguard will place some comments below the networking section of each node - to identify standby NICs.
EXAMPLE:

# Primary Network Interfaces on Bridged Net 1: lan3.
# Possible standby Network Interfaces on Bridged Net 1: lan0,lan2.

When Serviceguard creates the template file, it loads the NICs that it supports.
When cmapplyconf is performed, it determines which NIC are physically on the same network and labels that a bridged network.

cmviewconf will show which NICs are attached to which bridged networks.
EXAMPLE:
# cmviewconf | grep -e "Node name" -e lan -e bridge
Node name: helix
network interface name: lan3
bridged net ID: 1
network interface name: lan0
bridged net ID: 1
network interface name: lan2
bridged net ID: 1
Node name: torus
network interface name: lan2
bridged net ID: 1
network interface name: lan3
bridged net ID: 1
network interface name: lan0
bridged net ID: 1

In this case, there are 3 NICs on each server that are attached to the SAME physical network (because the bridged net ID is the same).

Also note, that like your example, this real example shows the NICs are not ordered the same on both nodes.

[melvyn burnard]

The order these lan cards appear in is irrelvant.
What is in your ascii file and what is actually useable can sometimes differ due to changes being madfe or failures.
If you run cmquerycl -C testascii -n -n take a look at the networking layout and see what comments are made, as per Steven's example.
Also, use cmscancl to get a scancl.out report and see what that shows regarding connectivity etc.
Finally, what does cmviewcl -v report, any LANs DOWN?

[manish_csc]

Thanks to all ,

My problem resolved...

[manish_csc]

closed