[SAC] Meeting Minutes

Folks,

The IRC meeting on SAC OSU OSL transition today was useful though our turnout
wasn't quite as high as I had hoped. The logs are at:

  http://logs.qgis.org/osgeo/%23osgeo.2010-06-10.log

The minutes at:

  http://wiki.osgeo.org/wiki/SAC:June2010_meeting#Minutes

The major decisions / take aways for me were:

  * We plan to complete migration off osgeo2 this month and shutdown by the
    end of June.
  * We plan to complete migration off osgeo1 by the end of September and
    shut it down.
  * We prioritized establishing a large "Projects" VM and the services for
    GRASS, Mapbender, and metadata.osgeo.org will be located their until there
    is a fairly compelling need for project specific VMs.

We still need to fine tune our backup strategy, timing, etc.

Some questions outstanding around the idea of doing all logins over https
to avoid making OSGeo passwords easily sniffed.

Best regards,
--
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up | Frank Warmerdam, warmerdam@pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush | Geospatial Programmer for Rent

Frank wrote:

* We prioritized establishing a large "Projects" VM and
the services for GRASS, Mapbender, and metadata.osgeo.org
will be located their until there is a fairly compelling
need for project specific VMs.

watching xblade14 for the last couple of weeks it seems that a
number of buildbots are responsible for a good portion of the
load. The run niced, but still.. So I've been thinking it might
be good to have a build.osgeo.org VM set up as a build server;
it could run at 100% without slowing down the static html front
ends of the other sites hosted on osgeo4. comments?

Some questions outstanding around the idea of doing all
logins over https to avoid making OSGeo passwords easily
sniffed.

a big +1 for adding https to the mediawiki login page(s), and
whatever ldap magic is needed to reuse the osgeo ids there.
..if we can avoid having a bad day..

regards,
Hamish

ps- continue this over at grass-dev, but re CMS for Grass I saw
www.lyx.org the other day and thought it looked good for our
'portal' front page needs. Aesthics are not too austere flat
text like trac wiki, not too blog or wiki like either. Actually
it uses PmWiki but the visitors don't know that, the login
button is very subtle. and they kept notes!
  http://www.lyx.org/SiteDocumentation
  http://www.pmwiki.org
convert html -> pmwiki
libhtml-wikiconverter-pmwiki-perl

No idea how it's stored on disk (eg if you can ues unix command
line power tools to bulk maintain pages) but it seems to me that
the home-portal doesn't really need to run more than about a
dozen pages of content- it's just a front end reception service.
No idea about RSS support, but we can't be the first folks to
think about it. Again, https logins and reuse of osgeo ids+
access groups would be a bonus.

On Thu, Jun 10, 2010 at 09:09:38PM -0400, Frank Warmerdam wrote:

The IRC meeting on SAC OSU OSL transition [...]

Thanks. I'm currently on the LinuxTag expo in Berlin, presenting
FlightGear and having just limited and flaky internet access. I'll get
back to this stuff next week when I'm back home,

  Martin.
--
Unix _IS_ user friendly - it's just selective about who its friends are !
--------------------------------------------------------------------------

Hamish wrote:

Frank wrote:

* We prioritized establishing a large "Projects" VM and
the services for GRASS, Mapbender, and metadata.osgeo.org
will be located their until there is a fairly compelling
need for project specific VMs.

watching xblade14 for the last couple of weeks it seems that a
number of buildbots are responsible for a good portion of the
load. The run niced, but still.. So I've been thinking it might
be good to have a build.osgeo.org VM set up as a build server;
it could run at 100% without slowing down the static html front
ends of the other sites hosted on osgeo4. comments?

Hamish,

We have such a server, and it is buildtest.osgeo.org - another
of the blades. Unfortunately there has not been manpower
resources to migrate all the buildbot slaves to that blade.
Some of the more critial/heavy ones have been moved - as I
understand.

Best regards,
--
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up | Frank Warmerdam, warmerdam@pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush | Geospatial Programmer for Rent

On Jun 11, 2010, at 9:11 AM, ext Frank Warmerdam wrote:

Hamish wrote:

Frank wrote:

* We prioritized establishing a large "Projects" VM and
the services for GRASS, Mapbender, and metadata.osgeo.org
will be located their until there is a fairly compelling
need for project specific VMs.

watching xblade14 for the last couple of weeks it seems that a
number of buildbots are responsible for a good portion of the
load. The run niced, but still.. So I've been thinking it might
be good to have a build.osgeo.org VM set up as a build server;
it could run at 100% without slowing down the static html front
ends of the other sites hosted on osgeo4. comments?

Hamish,

We have such a server, and it is buildtest.osgeo.org - another
of the blades. Unfortunately there has not been manpower
resources to migrate all the buildbot slaves to that blade.
Some of the more critial/heavy ones have been moved - as I
understand.

Also note that I'm not aware of any particular load issues on
xblade14; at least, we haven't had complaints of much other
than immediately after the qgis release (which is why we moved
qgis to a different host).

Are there ongoing performance issues on the xblade servers?

Regards,
--
Christopher Schmidt
Nokia