Miles O'Neal wrote:
> John Hearns said...
>
> |At the other end of the spectrum I would advise against 'everything'
> |installs. Been there, done that at a customers request for a 64 node V20
> |cluster. It takes a looooongggg time.
>
> That depends on your filers, network, and hardware.
>
> It also depends on your situation- how many nodes
> of this type will you install, how much testing can
> you do, how well do you know the packages, how often
> will you need to reinstall or add nodes, etc. If
> you know things inside out, you won't need as much
> testing. If you don't, you'll need more. If you
> re/install a lot, testing time is well spent. If
> you will install 64 nodes of this exact config and
> be done with it a while, then install 64 more, let
> it be a while, etc, it may be quicker just to do
> the install.
May I advise against an 'everything' install from a different
perspective, other than time.
DON"T DO AN EVERYTHING INSTALL ON ANYTHING PRODUCTION
I could be wrong, but I have yet to know anyone who actually does know
the in's and out's of every single package in Scientific Linux. This
includes me and Connie.
A good example is someone who complained that our LVM was broken. But
it seems they installed 'everything' which includes the cluster LVM, and
since they hasn't setup a cluster, LVM couldn't connect to the cluster,
so it wouldn't run correctly.
I do my best to save these 'everything' people from themselves, but
things get by.
Troy
--
__________________________________________________
Troy Dawson [log in to unmask] (630)840-6468
Fermilab ComputingDivision/CSS CSI Group
__________________________________________________
|