Warnings after IP changes on ServiceGuard ASCII (CONFIG) file

[Joe Kanakaraj]

Hi,

I am trying to configure a 2 node cluster. After running the cmquerycl command to generate the ASCII file. I notice a few things.

1. My stationary IP is my utp connection on a separate VLAN which we use specifically for heartbeats on the network. (Lan0)
2. My Heartbeat is IP is my fiber connection which is on the production VLAN. (Lan1)

Note: - Lan2 is my Stationary standby connection which is also fiber connected to my Production VLAN and is in unplumbed state.

Now when I run a cmcheckconf on the created file without editing, there are no warnings and it does the check clean.

After I go ahead and edit the ASCII file, and swap the IP’s and between my Heartbeat and Stationary IP I get the warning given below when I run the cmcheckconf on the edited ASCII file.


Gathering Network Configuration......... Done
Checking for inconsistencies .. Done
Warning: 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)

My question is why does this happen only when I swap the IP’s around, especially since I have just configured two similar setups with pretty much the same hardware and had no warnings. I do not have APA on the machine. I know I can form the cluster even with this warning, but I do not want to risk anything by taking it into production with it.

Insights, fixes and suggestions are what I need….

Cheers,

Joe

[Stephen Doud]

Please show a 'before' and 'after' look at your cluster configuration parameters (drop all of the comments except those related to the bridged networks). That is, what it looked like before you modified it, and after you "swapped" IPs.

Easiest way to gleen this info:
# grep -v ^# cluster.ascii | grep -v ^$

[Joe Kanakaraj]

Before Change Look....



CLUSTER_NAME cluster1
FIRST_CLUSTER_LOCK_VG /dev/vgSAN00
NODE_NAME lobster1
NETWORK_INTERFACE lan0
STATIONARY_IP 172.16.231.51
NETWORK_INTERFACE lan1
HEARTBEAT_IP 172.16.100.51
NETWORK_INTERFACE lan2
FIRST_CLUSTER_LOCK_PV /dev/dsk/c30t0d0
NODE_NAME lobster2
NETWORK_INTERFACE lan0
STATIONARY_IP 172.16.231.42
NETWORK_INTERFACE lan1
HEARTBEAT_IP 172.16.100.50
NETWORK_INTERFACE lan2
FIRST_CLUSTER_LOCK_PV /dev/dsk/c32t0d0
HEARTBEAT_INTERVAL 1000000
NODE_TIMEOUT 2000000
AUTO_START_TIMEOUT 600000000
NETWORK_POLLING_INTERVAL 2000000
NETWORK_FAILURE_DETECTION INOUT
MAX_CONFIGURED_PACKAGES 0
VOLUME_GROUP /dev/vgSAN00
VOLUME_GROUP /dev/vgSAN01



After Change Look..........



CLUSTER_NAME I-Link
FIRST_CLUSTER_LOCK_VG /dev/vgSAN00
NODE_NAME lobster1
NETWORK_INTERFACE lan0
HEARTBEAT_IP 172.16.231.51
NETWORK_INTERFACE lan1
STATIONARY_IP 172.16.100.51
NETWORK_INTERFACE lan2
FIRST_CLUSTER_LOCK_PV /dev/dsk/c30t0d0
NODE_NAME lobster2
NETWORK_INTERFACE lan0
HEARTBEAT_IP 172.16.231.42
NETWORK_INTERFACE lan1
STATIONARY_IP 172.16.100.50
NETWORK_INTERFACE lan2
FIRST_CLUSTER_LOCK_PV /dev/dsk/c32t0d0
HEARTBEAT_INTERVAL 1000000
NODE_TIMEOUT 2000000
AUTO_START_TIMEOUT 600000000
NETWORK_POLLING_INTERVAL 2000000
NETWORK_FAILURE_DETECTION INOUT
MAX_CONFIGURED_PACKAGES 0
VOLUME_GROUP /dev/vgSAN00
VOLUME_GROUP /dev/vgSAN01

[Mark Nieuwboer]

Hi Joe,

It's no problem, but serviceguard is for high availeble. it's normal to have a failover network card for your stationary and one for your heartbeat.
So you have to configure 4 cards. this is for the possibility a card will fail.
What you can do is to make everything heartbeat. So if your first heartbeat will fail your old stationary will take over.
The only negative about this it can cause a time out if there is a lot of netwok traffic over your fiber connection.

Hope it helped
grtz. Mark

[John Bigg]

With your initial configuration you have 1 lan with a stationary IP and no standby, in addition you have a heartbeat lan with a standby.

This meets the minimum requirements as:

1 heartbeat network with local switch

When you swap the addresses, if I understand what you are saying, you end up with 1 lan with a stationary IP and a standby, in addition you have a hearbeat lan with no standby. Therefore your heartbeat network has no protection and is a single point of failure resulting in the error message you are seeing.

I would strongly recommend that you run heartbeat over ALL lans since the overhead is very low and even with a standby card the lan itself is a single point of failure.

I'd suggest changing all occurances of STATIONARY_IP with HEARTBEAT_IP which will not only resolve your problem but will make the configuration more highly available.

[Stephen Doud]

Yup - change any reference of STATIONARY_IP to HEARTBEAT_IP to get redundant hearbeat capability. Such a designation does not prevent the NIC from carrying normal production data. It merely allows SG to have two paths to send HB packets.
That should eliminate the messages.