Tyler Mitchell (OSGeo) wrote:
Hi all,
Perhaps it's a good time to pause and ask a few questions that affect maintaining our services:
1- Do we have enough available SAC volunteers to help focus on these problems? Need to recruit a few more?
Tyler,
I think some additional volunteers would be helpful. I'm not sure
though how we know when we have too many and management is becoming
too chaotic.
2- Can we give a couple more people shell access to help? e.g. Folks from projects that have moved into OSGeo (GRASS, QGIS)
I think Chris Schmidt and Daniel Morissette are skilled admins
who could be an asset as primary administrators if they were
interested. Perhaps also folks from other projects.
I do however think that "primary administrators" need to make
some commitment to learning how things are setup and being available
to deal with stuff beyond just the one thing that they want to setup.
3- Could we use some hired help on-demand?
I think it would be helpful if there was on-demand hired help that
Howard could task to do stuff. I'm a bit leery about paying
commercial rates for this though as we could eat through quite a bit
of money quickly. My *hope* was that there might be a
consultant/contractor from somewhere with modest wage expectations, and
a passion for OSGeo that might take this on for a modest wage (say
$20/hr). Of course, I hope that for everything and it is hard to
get a solid mix of skills and willingness to work cheaply.
4- Do we want to consider moving to another hosting platform?
I'm not sure what you are proposing, but I think change without
a clear rationale is the last thing we need.
5- Should we move some services from one server to another?
This is plausible, but once again without a good understanding
of what the problem is, it seems like churn. We really need
someone decently skilled to watch the situation closely and try
to diagnose what is happening (perhaps more than one thing!).
This would include post-seizeup log analysis, adding various
sorts of instrumentation (logging server-status reports, top
to a log, etc) and then careful experimentation with variations.
It could take a while. I haven't observed any server problems
in the last several days for instance.
As with any volunteer group, we are dependent on goodwill and time availability of the committee and personal availability fluctuates over time. It might be a good idea for SAC members to comment on their current level of availability so we can see if there is a hole in support availability.
I am available as needed, but have rather adhoc system administration
skills. Doing stuff by ssh is very painful for me due to my extremely
poor latency internet connection so I feel like I should be a
resource-of-last-resort for most admin work. 
The hosting and service location questions are more what Arnulf seems concerned about. Even if we decide to migrate to another ISP or to move some services around on the boxes, I wonder if we have enough volunteer time available to make it happen effectively.
I'm not sure I follow this. I recall Arnulf mentioning website
(presumably www.osgeo.org Drupal) downtime and wiki downtime. I
don't recall all the details but I'm not sure either of these was
a problem with our ISP. One was mysql stuck with a broken table after
a hard reboot, and the other was mysql binary logs filling the disk -
a configuration error on our part.
I will say, both of these could have been either resolved more
quickly, or prevented in the first place by more ongoing volunteer
effort and more systematic management of our services.
All IMHO of course.
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 | President OSGeo, http://osgeo.org