[SAC] 2013 Agenda - was: Upgrade list for VMs

[Frank Warmerdam]

Alex,

Somehow I don't see the original message in this thread.

On Trac - I think it is suffering from robots requesting expensive
things (timelines over large ranges was one thing I saw).  I'll look
into that a bit more today and see if I can tune the robots.txt.

On "other items" - I'm wondering if we would be able to run some
windows VMs at OSU OSL?  I experimented a bit with using an Amazon
instance for windows builds, and it worked but it starts to get at
least a wee bit expensive to keen a reasonable fast and large instance
running constantly.  The downside to this is that it would require
somewhat different administration approaches than the set of similar
linux VMs.

Best regards,
Frank


On Thu, Jan 17, 2013 at 6:36 PM, Alex Mandel <tech_dev at wildintellect.com> wrote:
> On 12/13/2012 11:41 AM, Alex Mandel wrote:
>> Other things that have come up which should be addressed:
>> * Make sure all the machines are on Munin (missing - Secure, Mail, Web)
>> * Finish migrating osgeo1 to OSL servers (Drupal)
>> * Purchase of 3rd server for OSL - should it be slow with large disks,
>> or another VM machine to spread the load?
>> * Upgade Trac, we have a cloned server to test, if we're not going to do
>> it we should turn off that clone and recover the resources.
>> * Migrate Vmap0 to telescience
>>
>>
>> This is mostly a mental dump before I take off for a week of vacation
>> likely without internet.
>>
>> Thanks,
>> Alex
>
> Slight update:
> * We have a round-robin DNS with telescience for serving Vmap0 so we're
> not taking all the hits anymore. - thanks Martin
> * Progress is being made on Drupal migration from osgeo1, I've begun
> discussions with my coworker who is an expert on the topic and can
> probably solve this. I will check if spending funds might make this
> happen faster.
>
> * New Server
> I'm starting to think that we might want to get a new machine that is
> more VM capable, as trac seems to use as many cores as we can give it.
> Going this route, we could re-balance the VM load and free up more disk
> on osgeo4 for more backup space.
> I've restarted the conversation with OSUOSL since they made an offer to
> use a donation from us for a new machine (not sure that ever went through).
> Other plan, is just to get something with big disks, but I'm not sure
> we'd use that smartly over more cpu power for higher loads (likely to
> have quite a bit of disk anyways since price has come down so much)
>
> *Trac upgrade
> 1. We should move to mod_wsgi - I started looking at how complicated
> this would be, should be possible in place, 1 trac site at a time. (Easy)
> 2. We should upgrade trac, this can happen before or after adding server
> 3. It's easy to clone the running server, test the upgrade and then swap
> the 2 VMs ip addresses. (Harder)
>
>
> Anyone have comments or other topics to add?
>
> Thanks,
> Alex
> _______________________________________________
> Sac mailing list
> Sac at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/sac



-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Software Developer