[OSM-dev] planet.openstreetmap.org breakage

[Grant Slater]

Stefan de Konink wrote:
>> http://lists.openstreetmap.org/pipermail/talk/2009-March/035381.html
>
> So could you elaborate the downtime? We have a 'perfect history 
> system' and since we are talking 'all accounts have been identified' 
> wouldn't that be a perfect test case for our 'users that don't want to 
> migrate with the new license scheme'?

Downtime is limited to http://planet.openstreetmap.org/ while we patch 
Lithuanian data out of the old planet files.
This is not a test case and the revert tool is a special custom app for 
this single job.

> I could perfectly understand the need for removing the data content 
> from one user. Since our history system allows to 'delete' these edits 
> it will be a trivial task to do this and produce a new planet file. 
> After which you delete the history of the affected object (until the 
> point the user did not touch it). Of course doing this via the public 
> API might be a bit too slow (but hey this is also told to everyone 
> else that has a mass import or mass change). Then again a delete from 
> the current table in SQL matching the userid, is that going to affect 
> service; I doubt this.

This was not 1 account....
The revert was done via the public API. Looking up history for 
Lithuanian data was sql-hotwired.

>
> Therefore I don't get why downtime is required for a foundation core 
> service. (Especially looking at what date the notice was done... August)
> I do understand the need of removing all previous planets up to the 
> point of first data entry.

The original user was warned and blocked in August... He simple ran off 
and created sockpuppet accounts.
People, investigation, policy and apps take time.

Patched planet files will become available shortly, but they too take 
time to generate.

Regards
 Grant