[Linux-cluster] Diskless Shared-Root GFS/Cluster

[Hagmann, Michael]

When you come from the TruCluster side you will love it,like me.

But as Tom said is totaly diffrent and you should aware of, as example
Human Error like delete the /etc or others.
we implement for this reason ( as we do in our TruCluster Enviroment ) a
rsync based clone script. That mean we can clone with one Command the
/boot and root to a "clone /boot" and a "clone root". In the grub we add
a entry to boot the clone, when someone also delete the /boot, you have
to change the /boot disk in the qlogic menu to the clone /boot.
We also use this localclone, when we make a OS upgrade. First step is
make a localclone and then upgrade the system. When you have trouble
during the upgrade you always have the chance to boot the localclone and
you have a working cluster. It's also great when you have done the
upgrade and a few days later a big Problem ocoured, you can reboot into
the localclone and have a working cluster. 

What I wan't say is we are aware of this Problem ( because we use it
long time ) and we handle that with spez. Processes and Tools. The
Shared-Root it's a diffrent approach, like TruCluster and OpenVMS.

We love it

Mike

-----Original Message-----
From: linux-cluster-bounces at redhat.com
[mailto:linux-cluster-bounces at redhat.com] On Behalf Of Tom Mornini
Sent: Mittwoch, 31. Januar 2007 10:26
To: linux clustering
Subject: Re: [Linux-cluster] Diskless Shared-Root GFS/Cluster

We strongly considered this when implementing our cluster
infrastructure, but decided against it when we realized just how
devastating *any* problem with that shared root would be...

-- 
-- Tom Mornini, CTO
-- Engine Yard, Ruby on Rails Hosting
-- Reliability, Ease of Use, Scalability
-- (866) 518-YARD (9273)