[theforum] new server staus

[David Kaufman]

Hi Adrian,

"adrinux" <adrinux at gmail.com> wrote:
>
> David Kaufman <david at gigawatt.com> wrote:
>> where he says the version of MySQL we're currently running is too old
>> --damn.
>
> Actually IIRC there was further discussion on #evolt pointing out that
> MySQL 4.1 was available via debian-backports or somewhere - but since
> you'd supposedly bought a new server running etch it was a moot point.
> But then it turns out it wasn't running etch. Or ubuntu. And here we
> are :)

Arghh.  All project complications are caused by lack of communication!  I 
never chat on #evolt (or anywhere else).

If I had, I'd have pushed to upgrade MySQL on the current box and let the 
development go forward!

I coulda guessed that *my* part of this migration would take longer than a 
week, and then... Someone has to setup the DNS/bind server and migrate the 
data, MySQL 5, PHP 5, Drupal 6 have to be installed of course, and all the 
mail services (qmail, Mailman, SpamAssassin, ClamAV), I guess we're also 
running CVS (and/or svn), so those repositories have to be moved, possibly 
upgraded.

I don't mean to sound negative, but at my company we learnt along time ago, 
the hard way, that site redesigns (aka intense web development projects) 
should never be attempted in tandem with hosting, server, O/S and platform 
changes.  You gotta prioritize, and do one before the other.  When you're 
migrating, you need the site and its software to be a well-defined fully 
spec'ed and working target, so you can replicate that.  During intense 
development people tend to setup dev servers, staging sites, parallel 
copies of software, services and other scaffolding needed to support the 
development, and when your scaffolding is knocked over by cranes, it wastes 
time and focus and energy.  The syadmins have got 2 sets of requirements, 
keeping the old things up till the new things are ready, and simultaneously 
setting up all the new things that are needed.  Migrations like this are 
like moving a building, brick by brick, while the building is also being 
renovated, some walls torn down, new floors being added and pool installed 
on the roof :-)

Anyway, rant over.  I whined enough on this topic the last time we chose to 
launch a redesign, and change operating systems and hosting providers all 
at the same time (how badly did we miss that deadline, nine months?)

>> So I'll be burning the midnight oil tonight and if I can't get it done 
>> by
>> tomorrow (or if I "brick" the server :-))
>
> Hope things go, or went well - my timezone head is not working today :(

Actually GREAT NEWS there: I successfully booted the new server into debian 
etch at 4am this morning! (it's after lunch time as I write this)  so, 
about 6 hours ago... ~10am GMT?  Then I had to go to work.

I installed Debian etch on, and booted from, the *swap* partition of the 
CentOS server (no I'm not crazy -- that's actually how its done: 
http://www.underhanded.org/papers/debian-conversion/remotedeb.html).  And 
now that the big CentOS root partition is freed up, I can blow it away and 
copy my new root fs to it, remount everything the right way again and it'll 
be a bright, shiny new debian etch box!

Sorry I didn't get to Ubuntu in my virtual testing -- I must've done this 6 
or 7 times on my virtual etch and CentOS machines, getting it wrong in 
different and interesting ways each time, so that was the route I knew how 
to go.

Fortunately, I'm sure that, after all this, I'll be able to upgrade us to 
lenny, whenever it's released, with my eyes closed!

I'll finish this up tonight (er tomorrow morning to all you UKians out 
there) and then whoever is installing the web app servers, db servers, 
mailservers, dns servers, version control repositories, virus and spam 
defense systems can dive in!

-dave