[Systems] ping aslo master

[David Farning]

On Tue, Jan 19, 2010 at 10:16 AM, Aleksey Lim <alsroot at member.fsf.org> wrote:
> So, the major purpose for new boxes is clustering ASLO?

I was just throwing that idea 'out there' to see if it made sense.
The wins from my POV are that:
1.  RIT (CSH) has smart networking students looking for real world experience.
2.  ASLO and the technology we used to run and develop it are in
demand by employers.
3. As a private university RIT and Research Computing, is always
looking for corporate funding opportunities.

Open Source Labs at http://osuosl.org/ started when a student got
permission to host some of the Gentoo infrastructure on a spare
machine.  From there, the university started working with mozilla to
host and maintain their mirror network.  Bouncer was a OLS project.
Now it is a nationally know computing resource.

For Sugar Labs, the win is more distant and risky.  It is always
easier to scale up a service by putting it on a faster machine.  At
some point, the faster machine becomes a piece of expensive, esoteric
hardware.  But, we are nowhere near that point.

The value is in planting the seed with Research Computing and the CSH
to foster a project which can grow to become maintainers of this type
of web service.

The costs-
Additional time and labor setting up and racking several systems vs
just one machine.
Additional power consumption.
Additional maintenance keeping several used machines going vs one new machine.

If necessary, I will maintain the service on daily basis and remotely
mentor any student interested in working on the project.

david

> As I'm current ASLO code base maintainer I can pick this up.. but as a
> last option:) if there is no other willing people, since I'm not
> experienced in admin tasks and right now I'm busy w/ sugar releasing
> tasks.
>
> If there is people who is willing to support these tasks, you can ask me
> any question related to ASLO at any time on #sugar(nick alsroot). Also
> there is http://wiki.sugarlabs.org/go/Machine/Discovery_One project which
> David initiated to cluster ASLO.
>
> People who will settle adopting new boxes down, if installing ASLO on
> new boxes won't be very popular, please ping me once more and I'll do
> what I can.
>
> On Mon, Jan 18, 2010 at 05:24:11PM -0300, Bernie Innocenti wrote:
>> [cc += lfaraone, pfmeec, cjg, acj, systems@]
>>
>> Luke is in contact with the Wikimedia foundation for getting the
>> machines shipped once we get confirmation for the hosting.
>>
>> In addition to these machines, Alexander Jones, a RIT student, found us
>> yet another 1U box of similar specs available from the CSH.
>>
>> Summarizing:
>>
>>  3x  Dual   2.5GHz AMD,  3-4GB RAM, 1x80GB HD
>>  1x  Single 2.4GHz Xeon,   ??? RAM, 2x18GB SCSI HD
>>
>> Together, these boxes are likely to provide less processing power than
>> the dual-quad-core server we had in mind. Reliability would also be a
>> concern, but we probably need to make all components redundant anyway.
>>
>> On the positive side, we're going to save $3000 from our IT budget and
>> provide a valuable opportunity for collaboration with both Wikimedia and
>> CSH, as David points out.
>>
>> What remains to be seen, is who will physically rack the hardware and
>> perform the initial system installation and configuration. We already
>> have 4 virtual machine images that were configured and tuned during the
>> initial phase of clustering ASLO. We could hand them to a local sysadmin
>> who would just have to unpack them into a local filesystem and make them
>> boot. If it turns out to be difficult, we could install a fresh Ubuntu
>> Karmic system on top of which we'll rsync our filesystem images. Either
>> way, it may be a good challenge for sysadmin students.
>>
>> If we're lucky, we may end up with enough ethernet cards and (managed)
>> switch ports to setup a second LAN for private traffic such as database,
>> nfs and syslog. We'd also need enough storage to backup all these machines.
>>
>> If this scenario results in unacceptable overhead for RIT, we could
>> return to our original plan to get a new 2U machine shipped there.
>>
>> Thoughts?
>>
>>
>> On 01/18/10 16:34, Stephen Jacobs wrote:
>> > Bernie should know where we are currently at with RIT.
>> > Last I heard we were still waiting to hear about hosting confirmation.
>> >
>> > Would be very to see it happen with either RIT Research Computing or
>> > Computer Science House.  Research computing is more likely to be a more
>> > solid group to host with if it clears policy.  More of an official RIT
>> > tie and a 12 month per year facility.  Dunno how much of CSH activities
>> > die down over the sumer break.
>> >
>> > On Jan 18, 2010, at 1:44 PM, David Farning wrote:
>> >
>> >> CCing Bernie, and Prof. Jacobs.
>> >>
>> >> Hey Dogi, I have been focusing entirely on Ubuntu Sugar Remix for the
>> >> last 6 weeks. We really needed to get that project going in order to
>> >> establish vendor and oem relationships.  In order not to cause
>> >> additional friction with Redhat. I am laying low with the  RIT
>> >> project.
>> >>
>> >> I am fine with where ever you want to put ASLO.  I got a _little_
>> >> carried away splitting ASLO up in to separate machines.  It was a
>> >> fascinating technical problem with such an elegant solution.... which
>> >> ended up making everyone else's live harder.
>> >>
>> >> Just a thought, It looks like wikipedia is willing to donate some old
>> >> machines to host at RIT.  We could create some potentially valuable
>> >> partnerships by working with the RIT computer house and the networking
>> >> program to host ASLO on those machines at RIT.
>> >>
>> >> 1.  From a load POV if we off load ASLO, sunjammer can handle the
>> >> remaining load for at least several months....  This puts Bernie in
>> >> the position of acting proactive rather than reactively.
>> >>
>> >> 2.  We can probably create a relationship with Wikipedia to give
>> >> SL/RIT their old machines every year.  I understand their policy is to
>> >> replace machines every three years for space, power consumption, and
>> >> reliability reasons.
>> >>
>> >> 3.  I am not sure what the hosting policy at RIT is, but it would be
>> >> very cool if we could establish ASLO as signature project for the RIT
>> >> computer house.  If I remember correctly,  from my student days.
>> >> Getting real world experience on real world projects that we could put
>> >> on our resume was one of our primary goals.  Maintaining a system like
>> >> ALSO would qualify as that.
>> >>
>> >> Short term, this would be _much_ more work than sticking ASLO on a
>> >> faster machine.  But long term it could create some valuable
>> >> relationships and provide some students with some valuable experience.
>> >>
>> >> david
>> >>
>> >>
>> >> On Wed, Jan 13, 2010 at 2:33 PM, Stefan Unterhauser <dogi at sugarlabs.org> wrote:
>> >>> hi David,
>> >>>
>> >>> I m now already a month in Boston,
>> >>> my phone number is 617 909 2178,
>> >>> i restarted the VIGmeeting again ...
>> >>> but this time in #treehouse on OFTC
>> >>> http://me.etin.gs/treehouse/treehouse.minutes.20100113_1418.html
>> >>> as u can read there there hardware arriving
>> >>> what do u think?
>> >>> do u need help?
>> >>> please feel free to phone me
>>
>>
>> --
>>    // Bernie Innocenti - http://codewiz.org/
>>  \X/  Sugar Labs       - http://sugarlabs.org/
>> _______________________________________________
>> Systems mailing list
>> Systems at lists.sugarlabs.org
>> http://lists.sugarlabs.org/listinfo/systems
>>
>
> --
> Aleksey
>