redhat cluster configuration with multiple IP connections and bonding

Hi All,

I have a Cluster in a lab setup, the cluster config is based on CentOS 5.6 (all the latest updates).
At the moment it fails over with a single IP and has MySQL, Samba and Apache services. I have included a diagram hopefully showing the setup at the moment.

I need to add a new network configuration so that there is a bonded network (2 interfaces) which go to separate switches, this will be to provide full 100MB network connectivity to various sites around the UK.

This means that there will be 2 networks, one of these is bonded and connected to 2 separate network switches, the other is just a plain single network interface. Both setups need to have IP Failover working.

Has anyone got any ideas how I can configure this setup.


lab-cluster.png
LVL 19
joolsAsked:
Who is Participating?
 
jgiordanoConnect With a Mentor Commented:
If they are both on the same subnet this will be an issue, if not the intelligence is built in to say I want ip address 192.168.4.25 monitored and used. I have

eth0 is 192.168.3.1x on both nodes

bond0 is 192.168.4.1x on both nodes

It should use bond0 since it won't be able to actually bring the virtual IP up on eth0


0
 
jgiordanoCommented:
Do you still need to have the single network cable as a monitored resource?
0
 
joolsAuthor Commented:
Not entirely sure what you mean (learning RHCS as I go). Would all the Networks need to be montiored anyway?

The diagram shows the existing network connection, this is currently monitored.

I need to add to this an additional network which uses 2 nics connected to 2 switches and has a single IP Address, I believe this is done by bonding but it may require some additional configuration in RHCS.

We are currently using a Windows server which has team'd NIC's, the lab system is an attempt to have these clustered in a linux environment as the current storage configuration prohibits using a windows clustered setup.

0
Cloud Class® Course: Amazon Web Services - Basic

Are you thinking about creating an Amazon Web Services account for your business? Not sure where to start? In this course you’ll get an overview of the history of AWS and take a tour of their user interface.

 
jgiordanoCommented:
usually what I do is I set up 2 X bonded interfaces. The first bond0 is for cluster communication and the second bond1 is the actual monitored IP that will fail over.
0
 
joolsAuthor Commented:
So in my case, the monitored link will be the new bonded interface and the unmonitored interface will now be the existing link.

Is there any way of configuring 2 monitored links so I can have 2 ip fail over addresses?

0
 
joolsAuthor Commented:
How does clusters know which interface to assign the failover IP address to?

I cant see anything specific in the clusters.conf file.
0
 
joolsAuthor Commented:
OK, I just found out the monitored IP uses the interface releated to the FQDN so there can be only one monitored IP, in which case I'll have to use the bonded interface as the monitored link and just use the other interfaces for management stuff.

The networks are all on separate subnets so that wont be the issue.

I need to rethink the network side of things to make sure that the failover occurs on the correct network.

In the case of bonding, are there going to be any issues patching to 2 separate switches. The idea of this is to provide fault tolerance. I've never bonded interfaces before so I don't know if there are going to be any gotchas or if there is a specific bonding configuration I need to do (if there is such a thing).

0
 
joolsAuthor Commented:
I sort of managed to get the answer myself and have since had to delete the Cluster from the lab hardware. However, you did give some pointers.
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.