[gdal-dev] OGR FileGDB fails to copy data

[Even Rouault]

Le mardi 17 novembre 2015 23:36:29, David Vick a écrit :
> Hi,
> 
> I'm currently using OSGeo4W with ESRI's 1.4 version of the File Geodatabase
> API and trying to ingest a filegdb into postgresql 9.2.2 that is configured
> with ArcSDE and is using st_geometry as the default type.  When I execute
> an ogr2ogr -update -append on a feature class that contains either
> mulitlinestring or multipolygon I get an error that the copy statement
> 
> failed.  The debug info is listed below:
> >ogr2ogr -update -append -preserve_fid -fieldmap identity -f "PostgreSQL"
> 
> PG:"host=******* user=**** dbname=******* port=5432 password=*****"
> C:\Users\dvick99\Documents\fgcm\TDS61_Merged_FJ.gdb AeronauticSrf_AM
> --config PGSQL_OGR_FID "objectid" --config PG_USE_COPY "YES" --config
> PG_USE_BASE64 "YES" --config FGDB_BULK_LOAD "YES" --config
> PG_LIST_ALL_TABLES "YES" --config OGR_TRUNCATE "YES" --debug ON -progress
> 
> FileGDB: Really opening C:\Users\dvick99\Documents\fgcm\TDS61_Merged_FJ.gdb
> OGR: OGROpen(C:\Users\dvick99\Documents\fgcm\TDS61_Merged_FJ.gdb/03B3D220)
> succeeded as FileGDB.
> PG: DBName="fgcmdev3wk"
> PG: PostgreSQL version string : 'PostgreSQL 9.2.2 on
> x86_64-unknown-linux-gnu, compiled by gcc (GCC) 4.4.6 20110731 (Red Hat
> 4.4.6-3), 64-bit'
> PG: PostGIS version string : '2.0 USE_GEOS=1 USE_PROJ=1 USE_STATS=1'
> OGR: OGROpen(PG:host=lnsvr0800.gcsd.harris.com user=fgcm dbname=fgcmdev3wk
> port=5432 password=fgcm/054050E8) succeeded as PostgreSQL.
> PG: Using column 'objectid' as FID for table 'aeronauticsrf_am'
> PG: Field shape is of unknown format type st_geometry (type=st_geometry).
> 0...10...20...30...40...50...60...70...80...90...100 - done.
> PG: PQputCopyEnd()
> ERROR 1: COPY statement failed.
> ERROR:  ESRI: invalid hexa character in hexa string [30000]:
> CONTEXT:  COPY aeronauticsrf_am, line 1, column shape:
> "{C8CDC53F-9478-4CB9-9CD3-7ADA9558FFC2}"
> 
> OGR2OGR: 118 features written in layer 'aeronauticsrf_am'
> FileGDB: ref_count of C:\Users\dvick99\Documents\fgcm\TDS61_Merged_FJ.gdb =
> 0 now
> FileGDB: Really closing C:\Users\dvick99\Documents\fgcm\TDS61_Merged_FJ.gdb
> now
> 
> Does anyone have any other suggestions to try?  Our current process of
> loading data is taking in excess of 12 hours and getting ogr2ogr to work
> will improve this time to just minutes.

David,

Difficult to know what happens with just that.
Which GDAL version ?
You could try without PG_USE_COPY=YES to check if it is an issue with the copy 
mode.
And -skipfailures to identify if there's a particular feature that cause 
problem or if it is more general.
Perhaps a field mismatch with "-fieldmap identity" ? This option only works if 
the column names in source and target datasets are listed in same order (check 
with ogrinfo)
Did you try creating a new target layer instead of -update + --config 
OGR_TRUNCATE "YES"  ?

I'm not sure when the error "ERROR:  ESRI: invalid hexa character in hexa 
string [30000]:" comes from. Is there a 'ESRI' column ?

Even

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com