Quick note --
It seems there's been a long serious of miscommunications and dropped handling of access & software setup for the mapping servers, which has led to some very unhappy people. :(
I want to make sure we get this sorted out in the next few days... for the moment we're collecting notes on what actually still needs to get set up here:
http://meta.wikimedia.org/wiki/Maps_server_setup_tasks
Once we've got that down to a clear checklist and know who needs to do what I can bash people on the head until it happens. :)
-- brion
my Answers to some of the raised questions:
*Database server* 1. I'd suggest to start with only one read-only PostgreSQL server for both, the Toolserver & the WMF Rendering, but with each with it's own alias. Atm there's nearly nothing on the toolserver, but when the load produced by it grows, we could spawn a separate db-server, keeping the alias.
4. additional metadata like localized names can be included into the mapnik/PostGIS-DB without any problems, as its structure is thought to be customized this way. I don't know about other schemes.
*Rendering machinery* I'd suggest mapnik, as it's the one used on openstreetmap.org, so it's style is somehow the "face" of openstreetmap: this is what new people to osm think, osm looks like.
*Stylesheet management+ 1. Internationalization is well done using the views in the postgis-db, using the name:xx-Tags.
And some more Questions (I added these also to the article) *Main OSM mirror database* what will be mirrored? the current-tables the history-tables the raw-tables how could this be mirrored? only current can be imported from a planet.osm how often should this be updated?
*Mapnik database* we'll need a procedure for regular complete re-imports to solve inconsistencies introduced by the diff-import
Brion Vibber schrieb:
It seems there's been a long serious of miscommunications and dropped handling of access & software setup for the mapping servers, which has led to some very unhappy people. :(
We are all, by any means, only offering our private time to this. So it will take as long as it takes.
Peter
On 8/20/09 2:04 AM, Peter Körner wrote:
Brion Vibber schrieb:
It seems there's been a long serious of miscommunications and dropped handling of access& software setup for the mapping servers, which has led to some very unhappy people. :(
We are all, by any means, only offering our private time to this. So it will take as long as it takes.
Making sure whoever's doing the setup has the necessary *access* to the box shouldn't take long at all, which is why I want to get it going fast. :)
-- brion
Making sure whoever's doing the setup has the necessary *access* to the box shouldn't take long at all, which is why I want to get it going fast. :)
Aas far as i understand it's not the process of giving access, that takes long, but the process of deciding *who* is getting *which kind* of access to *which servers*. And I can understand that - I would also think twice before giving so. root-access...
But the time we're talking we're not able to work, so back to sth. practical.
Peter
Aas far as i understand it's not the process of giving access, that takes long, but the process of deciding *who* is getting *which kind* of access to *which servers*. And I can understand that - I would also think twice before giving so. root-access...
Folks, Brion is Wikimedia's CTO. He just told you that you just have to ask to get access. So, ask!
-- daniel
Folks, Brion is Wikimedia's CTO. He just told you that you just have to ask to get access. So, ask!
I don't want root-access as I don't have the knowlege on how to set up the osm-stuff. I have user-level access to cassini and for me as developer that's really, really cool!
But I saw some people resign, who once were willing to do the job, like Sven [1] or, in some way, Julian [2] or Aude [3].
As far as i can tell there are some people around that are willing to help, but the feedback they get is very weak. I think the main problem is giving root-access to someone unknown to wmf and this is perspicuous. But aude e.g. did not even ask for root-acces, just for the permissions to get postgres-user on cassini (see [4]).
To come to conclusion, are you, brion, willing to give sven, julian and aude the permissions they want and are you sven, julian and aude willing to help on setting up the systems? Then I hereby declare you man and wife. :)
Peter
[1] http://lists.wikimedia.org/pipermail/maps-l/2009-August/000189.html [2] http://lists.wikimedia.org/pipermail/maps-l/2009-July/000177.html [3] https://jira.toolserver.org/browse/TS-302 [4] http://lists.wikimedia.org/pipermail/maps-l/2009-August/000210.html
Still willing to help. ;)
Julian
On Fri, Aug 21, 2009 at 09:21, Peter Körner osm-lists@mazdermind.de wrote:
Folks, Brion is Wikimedia's CTO. He just told you that you just have to
ask to get access. So, ask!
I don't want root-access as I don't have the knowlege on how to set up the osm-stuff. I have user-level access to cassini and for me as developer that's really, really cool!
But I saw some people resign, who once were willing to do the job, like Sven [1] or, in some way, Julian [2] or Aude [3].
As far as i can tell there are some people around that are willing to help, but the feedback they get is very weak. I think the main problem is giving root-access to someone unknown to wmf and this is perspicuous. But aude e.g. did not even ask for root-acces, just for the permissions to get postgres-user on cassini (see [4]).
To come to conclusion, are you, brion, willing to give sven, julian and aude the permissions they want and are you sven, julian and aude willing to help on setting up the systems? Then I hereby declare you man and wife. :)
Peter
[1] http://lists.wikimedia.org/pipermail/maps-l/2009-August/000189.html [2] http://lists.wikimedia.org/pipermail/maps-l/2009-July/000177.html [3] https://jira.toolserver.org/browse/TS-302 [4] http://lists.wikimedia.org/pipermail/maps-l/2009-August/000210.html