[Systems] First weekly meeting and other matters...

[Bernie Innocenti]

On Wed, 2010-02-17 at 22:04 -0600, David Farning wrote:

> I have been chatting with luke..... Seems he is as surprised as the
> rest of us at the quantity they sent:)

Heh, server goodness...

I wish they sent 6 full-time sysadmins along with the machines :)


> The next hog, ASLO, should work out well at RIT.  Because the
> activities themselves are pushed through the CDN, ASLO is mainly cpu
> load.  I am happy with the hosting relationship at RIT because all
> parties are coming out ahead.  We have a cool infrastructure project
> which needs hosting.  RIT has geeks who need sysadmin experience.
> WIN-WIN. 

And geeks with sysadmin experience are now our main bottleneck imho. We
need dedicated maintainers for the various components of our
infrastructure, and especially:

 * http://bugs.sugarlabs.org/
 * http://planet.sugarlabs.org/
 * http://wiki.sugarlabs.org/
 * http://lists.sugarlabs.org/

Plus all the things we're not doing due to lack of manpower: voip, rt, a
nice main website done with Drupal or similar...

All of these services are relatively low traffi and could easily share
the same physical box (in fact, they do already :).
They can be easily maintained from remote (in fact, I'm doing it from
Paraguay :)


> Another option would be to put two or three machine in trans-world for
> redundantly hosting the wiki, web site, and email systems for Sugar
> Labs and Local Labs.

Soon or later we'll have to replace solarsail, our aging primary server.
However, this machine is of vital importance, so I'd rather not replace
it with old hardware which is *less* reliable:

bernie at solarsail:~$ uptime
 06:22:36 up 262 days, 23:03,  3 users,  load average: 2.25, 4.65, 4.13

Making all primary services fully redundant is probably not an option,
because not all services are as "easy" to cluster as ASLO.

-- 
   // Bernie Innocenti - http://codewiz.org/
 \X/  Sugar Labs       - http://sugarlabs.org/