[OSM-dev] Update on redaction bot and minutely diffs
toby.murray at gmail.com
Thu Jul 12 15:54:16 BST 2012
Andy just pointed out that it should probably be mentioned that you
only need to mess with the osmosis state file if your osmosis was up
to date with planet.osm.org when the invalid diffs were generated and
it progressed beyond sequence number 141272.
I believe Grant deleted the invalid diffs pretty quickly so if you
were trailing behind on updates or if you only update periodically,
your osmosis may not have gotten up to the invalid diffs and no action
is required. So check the sequence number before replacing the state
On Thu, Jul 12, 2012 at 9:32 AM, Toby Murray <toby.murray at gmail.com> wrote:
> As I mentioned yesterday, the bot caused some problems with minutely
> diffs. Andy just sent a more detailed message about the technical
> problems to the rebuild list but here is a quick update for the user
> side of things.
> Some invalid diffs were generated yesterday. These have been removed
> from planet.osm.org and replaced this morning with valid files after a
> workaround was put into osmosis. This means that if you have a setup
> that is consuming minutely diffs, you may need to manually intervene.
> According to Grant, the last valid diff was:
> So you will want to stop osmosis and replace its state.txt file with
> the contents of:
> And then restart your minutely update process. The first 20 or so
> diffs are substantially larger than normal so it may take a while to
> catch up.
> As for the bot itself, it was stopped after a few hours yesterday
> because some problems were discovered with its handling of relation
> members. Again, see the rebuild list for details. The short version is
> that he fixed it and as of a few minutes ago the bot is running again.
> There was also a problem caused in JOSM which might affect you if you
> are working in an area with other people. If someone else deletes a
> node and then you try to update it, JOSM had problems dealing with the
> deleted node because of a slight change in the API. This has already
> been fixed according to this trac ticket:
> So I think we've had our glitches. It'll be smooth sailing from here
> on, right? :)
More information about the dev