[OSM-dev] Known good version of planet.osm

[William Temperley]

I'm fairly sure the problem is random bit-flips when writing large files.

I've re-run the command:
osmosis --read-pbf-fast file=/pg_xlog/backup/planet-latest.osm.pbf
workers=4 --write-pgsql-dump directory=/user/osm

I ended up with junk between nodes 1884827207 and 1884827211 inclusive.

Then I ran the identical command, same file and I have managed to extract
the nodes that were previously destroyed.  There are no special characters
that may have caused UTF8 parsing issues and the nodes are all there intact.

Wow, this has cost me perhaps a week! Perhaps time to upgrade the RAID
array, which is well within support.

Best regards,

Will



On 31 March 2015 at 19:34, Paul Norman <penorman at mac.com> wrote:

> On 3/31/2015 4:13 AM, William Temperley wrote:
>
>> Dear all,
>>
>> I wonder if someone could point me to a recent version of planet.osm that
>> has been loaded successfully with Osmosis into the postgis snapshot schema,
>> please?
>>
>> The previous two versions (planet-150316.osm.bz2 and
>> planet-150323.osm.bz2) are giving me an error:
>> "org.apache.xerces.impl.io.MalformedByteSequenceException: Invalid byte
>> 2 of 4-byte UTF-8 sequence."
>>
> I'm downloading one of those to check, but are you sure it's a problem
> with the planet file and not Osmosis? Did you check the md5sum of what you
> downloaded?
>
> In either case, you should not be using the bzipped XML but should instead
> use a PBF, which is much faster to process.
>
> _______________________________________________
> dev mailing list
> dev at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20150410/cc23c8fc/attachment.html>