[Dachs-support] Spherepoly line segment overlap
Markus Demleitner
msdemlei at ari.uni-heidelberg.de
Wed Jan 3 13:51:39 CET 2018
Hi Sebastian,
[...and a happy new year to all DaCHS users...]
On Wed, Jan 03, 2018 at 11:34:29AM +0100, Sebastian Walter wrote:
> Dear all
>
> When trying to import my set of 3.700 images as WMS resources using "dachs
> imp q.rd" I get the following error message:
>
> > gavo imp -v q.rd
> > Making data import
> > Starting /var/gavo/inputs/fub_wms/data/wms.csv
> > Done /var/gavo/inputs/fub_wms/data/wms.csv, read 3700
> > *X*X* Uncaught exception at toplevel
> > *** Error: Oops. Unhandled exception InternalError.
> >
> > Exception payload: spherepoly_from_array: a line segment overlaps or
> > polygon too large LINE 1: ..., NULL, NULL, NULL, NULL, NULL, NULL,
> > NULL, spoly '{(2.96910...
> > ^
> Most probably a line segment overlaps as the data is often covering the
> poles (all polygons are defined by exactly 100 points).
...or it could be an issue with pgsphere; when you find the offending
polygon, would you send it to us so we can have a look at it?
> Is there a way to find out the CSV line number the error is appearing on? It
> is definitely not on line 1.
The reason DaCHS is unspecific here is that the problem happens when
a whole bunch of parsed rows has been passed on to postgres and
postgres chews on it. For performance reasons, DaCHS passes on 5000
records in one go by default, and the problem can be in any of them.
To pinpoint which record exactly bombs out, use the --batch-size (or
-b) option to tell DaCHS to just ship out one record at a time:
dachs imp -b 1 q.rd
When DaCHS fails after that, it is exactly at the row with the
polygon that makes pgsphere panic.
Add a -D to see what is actually passed to the database, which might
also help.
-- Markus
More information about the Dachs-support
mailing list