[OSM-dev] Daily diff's retry

[Brett Henderson]

Jon Bright wrote:
> Hi,
>
> spaetz wrote:
>   
>> The start time for the very first daily diff? Err, no idea it was a
>> manual run, which happened certainly after the planet dump started. But
>> usually the end time of the previous changeset is equal to the start
>> time to the next one. We could keep 8 of them around to make sure that
>> we always cover a whole week before a planet dump.
>>     
>
> So if I want to start using the daily diffs right now - should I apply 
> that very first diff?
>   
I asked Spaetz the wrong question above.  I should have asked what time 
was passed to osmosis in the intervalBegin argument to the 
--read-mysql-change task for the first run, not what time the process 
was actually launched.

1. If the intervalBegin argument contained a timestamp prior to the time 
the planet dump kicked off then you can start applying daily diffs now.
2. If the intervalBegin argument contained a timestamp after the time 
the planet dump kicked off then we can't apply daily diffs without some 
(probably minor) referential integrity issues.

If 1, then the answer to your question is yes, begin applying daily 
diffs beginning with the first diff.
If 2, then you have two options:
1. Apply it anyway, deal with any minor integrity issues that occur, and 
start again next week using the next planet when we have the required 
overlap of diffs and planets.
2. Ask Spaetz very nicely to manually create a changeset for the missing 
time interval (the whole previous day would be safest).

Hope that helps,

Cheers,
Brett