[OSM-dev] Updating Planet and Reliability

[Brett Henderson]

On Thu, Jul 21, 2011 at 6:18 PM, Frederik Ramm <frederik at remote.org> wrote:

> Hi,
>
>
> On 07/21/2011 10:07 AM, Andrew Ayre wrote:
>
>> What protection is there in Osmosis to recover from this without missing
>> any changes? If none, how are people solving this in their scripts?
>>
>
> I usually do this:
>
> * get latest changes, save in latest.osc
> * if a file named changes-to-apply.osc exists, merge latest.osc into that
> file; else rename latest.osc to changes-to-apply.osc
> * run osmosis to apply changes-to-apply.osc to my planet; remove
> changes-to-apply.osc if successful
>
> But it is possible that this is already over-cautious, and you could just
> as well run the --rri and --ac in one big step which would then ensure that
> the state.txt is only updated if the whole thing was successful. I'm not
> 100% sure about this though.
>

I think it's over cautious.  The --rri task won't update the state.txt until
the complete() call on the pipeline has returned.  If running within a
single threaded pipeline the --ac task and destination pbf (if that's what
you're using) task should be 100% complete before the state file gets
updated.

You will however need a wrapper shell script to rename the new planet file
to the original file name and this leaves a very small window for something
to go wrong between the state.txt being updated and the file being renamed.

Brett
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20110721/71a22842/attachment.html>