Re: [tlug] server installation best practices/ worksheet

[Al Hoang <hoanga@example.com>]

On Tue, May 22, 2007 at 05:27:10PM +0900, Keith Bawden wrote:

> >I was really just thinking that there should be some kind of generic
> >Linux checklist, or at least one for Red Hat systems. I will go ahead
> >and make one up myself, but it would have been nice to find one that
> >is used by folks more experienced than I.
> 
> If you have a fleet of boxes for a specific role then you would not
> likely find a generic checklist that fits your needs. It would be more

    Also, if you have a fleet of boxes it is well worth your time to
use a full-on configuration management system like cfengine, bcfg2,
puppet or whatever to handle your rollout.  For such a system,
the configurations files that specify how to make server FOO do BAR 
using BAZ in some ways becomes an executable checklist to get
some functionality out fo a box.

    Another option is to study how Santa moves around the planet
so fast in one night and emulate that :-)

> likely that you would create something specific to your use. If your
> system is truly generic then just go with a single large partition for
> everything minus swap. Then with swap go with 2 x RAM - that should
> get you by. Finally finish off with a Fedora or Ubuntu install
> selecting "server" as the type of system you are building.

    You might want to keep a seperate /boot partition if you are
still running yesteryears BIOSes :-)  I can't wait for such a blight
on the planet to end.

> 
> >However, I would really like to find best practices documents on
> >system and network documentation.
> 
> Best practices are a highly guarded set of secrets. Usually contained
> in either a set of books that are so numerous that several forests
> were sacrificed to source the paper, or they are locked away in the
> back room of someones head... The people who guard these will not tell
> you as they have forgotten why they decided they were best practices
> in the first place, or the system they were determined as best
> practices for is not longer in production and the new system is not
> compatible with the old set of best practices :-)

LOL.   On a more serious note.  My strategy is to try to find specific
documents on specific services and use that as a basis for building
a viable checklist.  Also keeping a diary of all the commands you
entered in to get some machine to do X is really useful in building a checklist.
Although it is very hard to remember to keep that diary.  Using script
or some similar tools to log your root / sudo sessions is really helpful
in this case.
    Upgrades are a double-edged sword in the systems
world as you can squash security bugs and gain much-needed (or not) features 
but end up having to toss out a few babies with the bathwater as newer
revisions on software you want to use always seem to find a way to goof 
up 'the way it used to work' in the name of 'simplifying the process'

> 
> Or, the best practices are so generic that they do not apply to your
> situation or any one else's ;-)
> 

    Unfortunately, when it comes to herding systems, "Works For Me"
tends to beats dusty, rusty, outdated, thoroughly documented checklist and
best practicies for the Linux distro that isn't even being distributed
anymore.  :-/


Alain