Single node cluster

[H_16]

Im running HPUX 11.23 Mc/SG 11.18

At the moment I only want a basic package for my NICs only..

Primary IP 172.168.2.143 (lan1)
Secondary IP 172.168.2.144 (lan5)
Virtual IP 172.168.2.145

I basically want if lan1 fails it will failover to lan5

Im new to MC/SG so can anyone tell me the steps are require in being a package with these details

[Mel Burslan]

In /etc/cmcluster directory, you should have an ascii cluster definition file, more than likely, named as cmclconf.ascii. When you edit this file, you will se a construct looking like this:


NODE_NAME sunshine
NETWORK_INTERFACE lan1
HEARTBEAT_IP 10.22.21.8
NETWORK_INTERFACE lan3
NETWORK_INTERFACE lan2
HEARTBEAT_IP 10.14.21.8
NETWORK_INTERFACE lan6


In this example here, lan1 is the primary interface with an IP address od 10.22.21.8 and if it fails, this IP address is going to be taken over by lan3. In the same train of thought, lan2, another interface needed by my package, has an IP address of 10.14.21.8. If this one fails for some reason, IP address will be taken over by lan6.

Hope this makes it a little clearer for you. But you need to understand that, if the only reason you are setting up a single node cluster, is to have network redundancy, please reconsider using HP Auto Port Aggregation (HP-APA). It is a much better way of doing it.

HTH

[Rita C Workman]

I copied this from somewhere and used when we had to do a slightly different thing. Break up an 8 node cluster into single node clusters for DR purposes. Along with Mel's instruction to NOT assign an IP to your second failover lan, you might this info helpful too.

Managing a Single-Node Cluster
The number of nodes you will need for your Serviceguard cluster depends
on the processing requirements of the applications you want to protect.
You may want to configure a single-node cluster to take advantage of
Serviceguardâ s network failure protection.
In a single-node cluster, a cluster lock is not required, since there is no
other node in the cluster. The output from the cmquerycl command
omits the cluster lock information area if there is only one node.
You still need to have redundant networks, but you do not need to specify
any heartbeat LANs, since there is no other node to send heartbeats to.
In the cluster configuration ASCII file, specify all LANs that you want
Serviceguard to monitor. For LANs that already have IP addresses,
specify them with the STATIONARY_IP keyword, rather than the
HEARTBEAT_IP keyword. For standby LANs, all that is required is the
NETWORK_INTERFACE keyword with the LAN device name.
Single-Node Operation
Single-node operation occurs in a single-node cluster or in a multi-node
cluster, following a situation where all but one node has failed, or where
you have shut down all but one node, which will probably have
applications running. As long as the Serviceguard daemon cmcld is
active, other nodes can re-join the cluster at a later time.
If the Serviceguard daemon fails when in single-node operation, it will
leave the single node up and your applications running. This is different
from the loss of the Serviceguard daemon in a multi-node cluster, which
halts the node with a TOC, and causes packages to be switched to
adoptive nodes.
It is not necessary to halt the single node in this scenario, since the
application is still running, and no other node is currently available for
package switching.

However, you should not try to restart Serviceguard, since data
corruption might occur if the node were to attempt to start up a new
instance of the application that is still running on the node. Instead of
restarting the cluster, choose an appropriate time to shutdown and
reboot the node, which will allow the applications to shut down and then
permit Serviceguard to restart the cluster after rebooting.


Rgrds,
Rita

[Duncan Edmonstone]

You are missing a basic point of the way SG LAN failover works here

If you want SG to just do LAN failover for you and nothing else, you don't need a package at all, or a virtual IP. However the standby LAN card has to have no IP bound to it, so for starters get rid of that extra IP address on lan5 with:

ifconfig lan5 0

Now in your case if you simply want an IP address which is normally on lan1 and then moves to lan5 if lan1 fails, you really only need the following in a cluster configuration file:

CLUSTER_NAME mycluster
NODE_NAME
NETWORK_INTERFACE lan1
HEARTBEAT_IP 172.168.2.143 NETWORK_INTERFACE lan5
HEARTBEAT_INTERVAL 1000000
NODE_TIMEOUT 2000000
AUTO_START_TIMEOUT 600000000
NETWORK_POLLING_INTERVAL 2000000
MAX_CONFIGURED_PACKAGES 10

Check that with "cmcheckconf -C "
and then apply it with "cmapplyconf -C "

Then start your cluster with cmruncl

You will now find that the IP address on lan1 of 172.168.2.143 will failover to lan5 if you pull the cable from lan1

HTH

Duncan

[Duncan Edmonstone · Accepted solution]

hmmm formatting got a bit screwed up there - that file should look like this:

CLUSTER_NAME mycluster
NODE_NAME
NETWORK_INTERFACE lan1
HEARTBEAT_IP 172.168.2.143
NETWORK_INTERFACE lan5
HEARTBEAT_INTERVAL 1000000
NODE_TIMEOUT 2000000
AUTO_START_TIMEOUT 600000000
NETWORK_POLLING_INTERVAL 2000000
MAX_CONFIGURED_PACKAGES 10


HTH

Duncan

[Viktor Balogh]

as Rite wrote it there's no sense of a HEARTBEAT_IP in a single-node cluster, instead of it i would write STATIONARY_IP.
otherwise it's ok.

[H_16]

I have just installed MC/SG A.11.18

There doesnt seem to be a cmclconf.ascii file...or do I have to create myself by running specific commands?

[Ganesan R]

Hi,

You have to create the cluster ascii file by running this command.

#cmquerycl -v -n nodename -C cmclconf.ascii

[H_16]

Should my netconf on have lan1 configured..
I have commented out lan5

INTERFACE_NAME[1]="lan1"
IP_ADDRESS[1]="10.124.3.144"
SUBNET_MASK[1]="255.255.255.0"
BROADCAST_ADDRESS[1]=""
INTERFACE_STATE[1]=""
DHCP_ENABLE[1]=0
INTERFACE_MODULES[1]=""

#INTERFACE_NAME[2]="lan5"
#IP_ADDRESS[2]="10.124.3.145"
#SUBNET_MASK[2]="255.255.255.0"
#BROADCAST_ADDRESS[2]=""
#INTERFACE_STATE[2]=""
#DHCP_ENABLE[2]=0
#INTERFACE_MODULES[2]=""

[H_16]

should have read...

Only have lan1 configured..

[Ganesan R]

Hi,

Yes. lan5 should not have any IPaddress bound to it.

You need to specify your virtual IP inside your package configuration file.

If you need to have another virutal IP you should configure another package

[H_16]

I have this configuration in my

CLUSTER_NAME mycluster
NODE_NAME
NETWORK_INTERFACE lan1
HEARTBEAT_IP 172.168.2.144
NETWORK_INTERFACE lan5
HEARTBEAT_INTERVAL 1000000
NODE_TIMEOUT 2000000
AUTO_START_TIMEOUT 600000000
NETWORK_POLLING_INTERVAL 2000000
MAX_CONFIGURED_PACKAGES 10

Where do I specify my virtual IP ??

I also run cmcheckconf -C xxx.ascii and the following was issued..


Note : a NODE_TIMEOUT value of 2000000 was found in line 119. This value
is recommended if the top priority is to reform the cluster as fast
as possible in case of failure. If the top priority is to minimize
reformations, consider using a higher setting. For more information see
the cluster configuration ASCII file or the Managing Serviceguard manual.
Network interface lan5 on node nodexxx couldn't talk to itself.
lan5 on node nmesito1 cannot be configured in the cluster
because it does not have an IP address, and it is not a standby lan for any other lan.
Failed to evaluate network
cmcheckconf: Unable to reconcile configuration file xxx.ascii
with discovered configuration information.

[Duncan Edmonstone]

Do you need a virtual IP? Once setup the IP which is currently on lan1 will move to lan5...

However right now you have another problem... from the output to the cmcheckconf/cmapplyconf commands you gace it appears that lan1 and lan5 can't talk to each other... are they plugged into the same broadcast domain?

HTH

Duncan

[Rita C Workman]

Hello H

First you obviously need to either go to an HP MC/SG class that HP offers (wwww.itrc.hp.com/education) or you need to download and ready the manual on this (www.docs.hp.com). With that said, here's basic steps for you:
==========================================
1. Get that IP of lan 5-here's 1 way:
ifconfig lan5 unplumb
2. On your single node server:
cd /etc/cmcluster
vi cmclnodelist
Put in:
root
Save the file!!
3.cmquerycl â k â v â n -C .ascii

This will create a cluster ascii or config file template to review.
Review this file and edit accordingly for cluster on this single node;
Add CLUSTER_NAME
Lock Disk info not required.
NODE_NAME & Network connections reflected
Change NODE_TIMEOUT to 100 00 00 0 (no space b/t zeros)
AUTO_START_TIMEOUT 600 00 00 00 â
NETWORK_POLLING_INTERVAL 200 00 00 â
MAX_CONFIGURED_PACKAGES <#ofpkgs>

-- Now I put spaces in those so you could tell how many to type, but do not put spaces in when you edit the file!
Save the file.
4. cmcheckconf -k -v -C .ascii
When clean run:
5. cmapplyconf -k -v -C .ascii
6. cmviewcl -v
=========================================

To respond to the errors you are currently seeing:

1. NodeTimeOut - covered above use that value.
2. lan5 can't talk to itself.
Check the lan/nic by running:
lanscan (to get MAC address of lan5)
linkloop (whole MAC addr)
If it comes back OK then lan card is fine. If it comes back with error, then check the cable, check the port on switch or lastly your card may not be working!
3. Virtual IP - are you trying to run this on a Virtual Machine? Stick with the basics first.


Rgrds,
Rita

[Rita C Workman]

I did a cut&paste from my notes here, so where you see: â Â
just put a dash....

/rcw

[H_16]

I have carried out the steps you have suggested.

when running the cmcheckconf command, I am receiving errors about lan5.

So as Rita suggested on LAN5
#lanscan
#linkloop 0x001083F74A4D
Link connectivity to LAN station: 0x001083F74A4D
error: expected primitive 0x30, got DL_ERROR_ACK
dl_error_primitive = 0x2d
dl_errno = 0x04
dl_unix_errno = 57
error - did not receive data part of message



However I did the same linkloop command for my LAN1 card and received the same error as above? any ideas....because my lan1 card is working as my server is on the network??

[Duncan Edmonstone]

you are not using linkloop correctly... the point is that if the 2 NICs are in the same broadcast domain, then you should be able to lonkloop between them, but you need to tell linkloop which lan card to send data out of using the "-i" argument. So you should be able to do the following without error:

linkloop -i 1

and

linkloop -i 5

if you can't do that without error, then taken in concert with the error you gotr from cmcheckconf/cmapplyconf we can deduce that the 2 lan NICs are not connected to the same broadcast domain - by which I mean:

i) in the same VLAN

AND
(
ii) connected to the same switch

OR

iii) connected to a different switch, but with spanning tree between switches
)

HTH

Duncan

[Jozef_Novak]

Hello,

why not to use APA to make this working ? It offers exactly what you want to have and with 11iv3 it is included with base OS.

J.

[H_16]

Duncan,

I am still receiving errors from both linkloop commands..

My network team has confirmed we have

iii) connected to a different switch, but with spanning tree between switches

Any other ideas??

[Duncan Edmonstone]

> My network team has confirmed we have
>
> iii) connected to a different switch, but
> with spanning tree between switches

Well it ain't working is it!

If you can't linkloop from one NIC to the other, then the NICs are *not* in the same broadcast domain.

Do they do anything funny with VLAN tagging in your network? (For instance tagging packets based on protocol or based on IP address) Some networks that don't have port based VLANs, won't necessarily broadcast a packet if it's not an IP packet or not in a specific IP subnet (which linkloop packets won't be)

As for the comment about APA - I think we already established you don't have that as you are on 11iv2, not 11iv3 - APA would have exactly the same challenges anyway...

HTH

Duncan

[H_16]

Thanks for all the help....the card was upside down so I should have configured lan8...doh!