[volunteers] brie

[Heather Stern]

On Mon, Apr 23, 2007 at 06:02:35PM -0700, Paul Reiber wrote:
> Hello there!
> 
> I learned from a phone call with Heather that she now has brie, and
> that Daniel's efforts to get it booting have worked.
 
Actually, it's more like - Daniel believs that he's fixed it so it would now
come up fine if it gets attached back to the JBOD;  it is still
interdependent on it.

If you want it back immediately give me a call and you can come get it.
However if it's ok with you I'd rather give it more of a looksee since it's
already detached and here.

> So... if that's really the case that's great news!
> 
> Daniel - can you tell us, what was wrong with it?
 
To keep it simple, GRUB got a mite confused.

> Heather - we should coordinate getting brie back to me so I can get it
> reinstalled at via.net -
> are you around Palo Alto or Mountain View during the day? Or sometime this 
> week?

I work days.  Also I work pretty hard;  the time I plan to spend on brie
should be short (and mildly noisy; its fans aren't terribly quiet unless you
compare it against other rack systems) but has to be during time that I'm
safe and sane at a root prompt and not intermingled with $paidstuff.   
 
> If you'd like to hold onto it for a while, while you're working with
> it (Moin! Woo Hoo!) just say the word & I'll wait for an "OK, I'm
> done; you can have it now" email.

Yes, definitely.  The list of TODO is brief but tasty:
  * RAID config smarter setup + notes in /etc/sysadmin/README.brie
    -> avoid: hang on startup due to RAID outages or workbench time
  * delink swaps so it's two of them => more vmem system bang for the buck
  * delink root fs => alternate root available quick rather than depend on
                      RAID for it if RAID itself goes sour
    + including basic "obvious" sysadmin tools like less and vim..
    + faster than finding a RAID capable liveCD in a hurry, or at least
      only requiring one that can chroot you into this partition and whose
      kernel isn't a total loss...

    ::: I know it had to come out this time but I'd rather make us able 
    ::: to deal with it fast onsite the next time some tidbit of hardware 
    ::: goes kerblooey.  Darn thing's *heavy*.

  * examine what it's trying to do for firewalling/service defenses
    + first with a mind to knowing/finding what they already are
      + Daniel told me what he considers some of the defenses
      + other comments/preferences welcome *please* speak up if you
        have a strong opinion on this!
    + second to plan them for a service-defends-itself view
      e.g. cymru style config + chroot for DNS, smart settings for apache
    + third so we're all agreed on what they ought to be before really 
      nailing it down and shoving it back in the cage
    -> avoid: locking ourselves out of our own box.

    ::: I don't want it going back into a colo in any condition that
    ::: we might have difficulty accessing remotely after it's in.

  * iff you folks are interested I can check the imprinted version on the
    motherboard and see if it can take more memory than it presently owns;
    iff the net says that's not a BigBadIdeaTM (I *have* killed mb by giving
    them too much RAM) -and- I happen to have the called for RAM type on
    hand (fairly likely; I collect spare chips every netlounge run; Rick's a
    bit of a collector too) I could then upgrade it.  Opinions (y/n) welcome.

    ::: I think 512M is ok but it might be nice if it could do better -
    ::: This is -*NOT*- critical.

  * which services are supposed to be on this thing? I can checklist through 
    them all while I've got it.
  * proper python, current moin, yes Rick I will make sure there's a bloody
    easy and obvious upgrade path.  /usr/bin/update-moin possibly :)
  * Other than obviously mailman* did we want other webby apps in?

  ::: it was out for hardware but I may as well check software status too.

         * god that thing's ugly.  I might at least adjust its coloration
           to match the rest of the site.  Properly skinning mailman is
           beyond the scope of a brief system checkup by Dr.Starshine :)
           If Lisa wants to give it a shot and succeeds I'll buy her 
           dinner sometime. 
 
Good weather willin' and the creek don' rise it may be feasible to get it
back in the cage this weekend*.  Unless you enjoy midnight rides to a colo
I don't think you'll get it back midweek even once I have it peachy-keen.
When I say I work days - I mean it, the daylight's booked, twilight time's
for dinner and by then I do need it, so nothing gets between geek and her 
mocha with mealtime :D

         * As usual much of my weekend is doublebooked and various efforts
           want my time at various meetings; just let me know which times
           work at all, and I'll figure out which group loses the musical
           chairs this time around. 

In the interests of sanity (and my crammed lab space) I'll be happy to hand
it off as soon as I feel confident it will remain remotely accessible when
locked back in.

> Thanks everybody!
> -Paul

Hey, it's fun, and needs doing :)

  . | .    Heather Stern          |     (408) 374-7623 land
--->*<---  star at starshine.org   - * -   (408) 761-4912 cell
  ' | `    KG6ZYC                 |