Re: failSAFE IP - Looks good! Who's using it? (and why/why-not)

From: Dave Harrold (DHarrold_at_wi.rr.com)
Date: 11/10/05 

Date: Thu, 10 Nov 2005 09:35:04 -0600

On Thu, 10 Nov 2005 13:19:55 +0800, "Richard Maher"
<maher_rj@hotspamnotmail.com> wrote:

>Hi,
>
>I came across this "failSAFE IP" documentation the other day and became
>quite interested in the functionality that it provides. My brief
>understanding is that when an interface (or whole other node in a cluster)
>dies it can automatically ifconfig its IP address to another interface in
>the same box (or other box in the cluster). And when the original interface
>becomes available again then it fails back seemlessly-like. Does that sound
>about right?
>
>Leaving aside cluster-aliases and DNS load-balancing servers and all the
>other good IP-clustering stuff for the moment, is anyone out there using
>failSAFE? Any brick-bats or bouquets?

Yes, we are using it in our production cluster. We have 2 interfaces
into the public side of our network, each to a different switch. By
setting up failsafe IP, we have been able to survive various netowrk
maintenance activities by maunally failing over the interfaces to the
still working network path.

We have also survived various failures (hate when someone pulls the
wrong network cable out) with out any loss of service.

Very cool stuff!

One problem we ran in to was the use of the dedicated lock manager. We
use the lock manager on out application nodes and the locking done to
make failsafe IP work had a problem with it. Basically, it would move
the address, but you would never be able to move it back. If you use
the dedicated lock manager, there is a fix available. If you don't,
never mind. :-)

>
>You're normally all over this sort of stuff Kerry; thumbs up or down?
>
>Regards Richard Maher
>

Hope that helps,

Dave Harrold

..............................................................................
David Harrold E-Mail: David.Harrold@aurora.org
Lead Software Systems Engineer

Aurora Health Care
3031 W. Montana Street
Milwaukee, WI 53215

Relevant Pages

  • Re: [PATCH 0/7] dlm: overview
    ... > aren't just unique within a single cluster (think clusters of clusters, ... How the configuration gets from the config file to kernel is a mystery to me ... By a message over a socket, ... Let's have no magical filesystems in the core interface please. ...
    (Linux-Kernel)
  • Re: Strange HACMP config error
    ... > had a full network blackout yesterday. ... The cluster was instable after ... > First one IP interface is gone. ... network switch logs can show routing problems if you have access to them. ...
    (comp.unix.aix)
  • Re: iscsi multipath fails when cluster service is started
    ... i've configured a virutal machine with windows server 2003 under esx4 to ... test the behavior of the network inferfaces. ... when i do the same procedure on my physical server with installed cluster ... interface still has a connection. ...
    (microsoft.public.windows.server.clustering)
  • Cluster Help?
    ... I am having an issue with my second node of an active/passive cluster. ... in cluster administrator both of the network cards are ... contexts for node 1. ... interface 1f387670-01f2-4fb1-b728-95b03d9321c2 ...
    (microsoft.public.windows.server.clustering)
  • Re: Failover on public network failiure?
    ... There are a number of ways the cluster service checks the state of the ... If the network interface is determined to be failed then any ... the IP address on the current node the groupwill failover to another ...
    (microsoft.public.windows.server.clustering)