[postgis-users] crash on ST_Clip()

Thu Apr 26 10:58:26 PDT 2012

Steve,

You'll need to use r9649 or later.  The error messages you're seeing in
the log aren't directly causing the segfault.  The first message has
nothing to do with the segfault and I have no ideas where that's coming
from except that your installed PostGIS SQL API is out-of-date.  The
second occurs along the way to the code that caused the segfault.

-bborie

On 04/26/2012 10:55 AM, Stephen Crawford wrote:
> Bborie,
> 
> So are you saying these are minor errors and are not causing the crash? 
> Or do you think it could be causing the crash and I need to get r9649?
> 
> Thanks,
> Steve
> 
> On 4/26/2012 12:30 PM, Bborie Park wrote:
>> Steve,
>>
>> The o_column column doesn't exist in 2.0 but did at one time prior to
>> that.  I'd guess you need to update the SQL API using
>> rtpostgis_upgrade_20_minor.sql.
>>
>> As for your second error, I believe that that error message is being
>> kicked back from GDAL BUT it isn't the actual error that is causing the
>> segfault.  Please look at ticket #1782 as I think you're hitting that
>> bug.  I fixed that bug in r9649.  You will want to try that revision
>> instead.
>>
>> GDAL is correct when it responds saying that EPSG:105002 is not found as
>> 105002 isn't a valid EPSG (epsg-registry.org).
>>
>> -bborie
>>
>> On 04/26/2012 08:50 AM, Stephen Crawford wrote:
>>> OK, before I try to assemble all the data, a couple more questions.  I
>>> was finally able to access the logs and I see two errors:
>>>
>>> LOG:  database system is ready to accept connections
>>> LOG:  autovacuum launcher started
>>> ERROR:  column "o_column" does not exist at character 39
>>> STATEMENT:  select o_table_name, overview_factor, o_column,
>>> o_table_schema from raster_overviews where r_table_schema = 'public' and
>>> r_table_name =
>>> 'spring_ms' and r_column = 'rast'
>>> ERROR 6: EPSG PCS/GCS code 105002 not found in EPSG support files.  Is
>>> this a valid
>>> EPSG coordinate system?
>>> LOG:  server process (PID 5497) was terminated by signal 11:
>>> Segmentation fault
>>> LOG:  terminating any other active server processes
>>> WARNING:  terminating connection because of crash of another server
>>> process
>>> DETAIL:  The postmaster has commanded this server process to roll back
>>> the current transaction and exit, because another server process exited
>>> abnormally and possibly corrupted shared memory.
>>>
>>>
>>>
>>> For the first error, there is indeed no column "o_column" in the
>>> raster_overviews table.  There are also no records in the table it at
>>> his point.
>>>
>>> For the second error "code 105002 not found in EPSG support files",
>>> this projection is one I created to match the grids we use.  I inserted
>>> the definition in spatial_ref_sys.  It is also listed in the
>>> raster_columns and geometry_columns tables  Is there somewhere else I
>>> need to define it?  I have used this projection in postgis for at least
>>> 4 years now.
>>>
>>> I also noted that topology is not installed, if that matters at all.
>>>
>>> postgres 8.4.9
>>>
>>>
>>> postgis_full_version
>>> ----------------------------------------------------------------------------------------------------------------------------------------
>>>
>>>
>>> POSTGIS="2.0.0 r9605" GEOS="3.3.2-CAPI-1.7.2" PROJ="Rel. 4.7.1, 23
>>> September 2009" GDAL="GDAL 1.9.0, released 2011/12/29" LIBXML="2.7.6"
>>> LIBJSON="UNKNOWN" RASTER
>>>
>>> Thanks again,
>>> Steve
>>>
>>>
>>>
>>> On 4/25/2012 5:05 PM, Paul Ramsey wrote:
>>>> In order to fix a problem, developers must reproduce it. What do we
>>>> need to reproduce it: the offending SQL and the data itself. In ideal
>>>> test cases the users can embed the data statically into a single SQL
>>>> statement. In others they provide minimal versions of the offending
>>>> tables. Try to make reproducing the problem as easy as possible for
>>>> others, and the odds of it being fixed go up. Create a ticket with the
>>>> SQL and description and attach your data to it.
>>>> <http://trac.osgeo.org/postgis>
>>>>
>>>> P.
>>>>
>>>> On Wed, Apr 25, 2012 at 1:29 PM, Stephen Crawford<src176 at psu.edu>  
>>>> wrote:
>>>>> Well I tried the same query for just one of the tiles that I know
>>>>> intersects
>>>>> the the_geom but it crashed.  Then i simplified it further by using
>>>>> just a
>>>>> box instead of a multipolygon and it still crashed.
>>>>>
>>>>>
>>>>> On 4/25/2012 3:22 PM, Paul Ramsey wrote:
>>>>>> Find the exact geometry pair that is causing the crash and file a
>>>>>> ticket. (Yes, disconnecting means the backend went boom, that's a
>>>>>> crash).
>>>>>> P
>>>>>>
>>>>>> On Wed, Apr 25, 2012 at 12:15 PM, Stephen
>>>>>> Crawford<src176 at psu.edu>     wrote:
>>>>>>> All,
>>>>>>>
>>>>>>> I have a query which works on my desktop Windows (postgis2.0 RC) but
>>>>>>> crashes
>>>>>>> on Linux (postgis2.0 official release).  And by crash i mean
>>>>>>> disconnects
>>>>>>> me
>>>>>>> from the server.
>>>>>>>
>>>>>>> INSERT INTO result_raster_table (observation_date, rast)
>>>>>>> SELECT r.observation_date, ST_Clip(r.rast,g.the_geom,-9999) AS rast
>>>>>>> FROM raster_table r, geom_table g
>>>>>>> WHERE r.observation_date = '2012-03-09';
>>>>>>>
>>>>>>> 'raster_table ' holds tiles, 187 of which make up one
>>>>>>> observation_date,
>>>>>>> and
>>>>>>> 'geom_table' is just one record, the union of ~23 Eastern US states.
>>>>>>>
>>>>>>> Any thoughts?
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Steve
>>>>>>>
>>>>>>> -- 
>>>>>>> Stephen Crawford
>>>>>>> Center for Environmental Informatics
>>>>>>> The Pennsylvania State University
>>>>>>> src176 at psu.edu
>>>>>>> 814.865.9905
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> postgis-users mailing list
>>>>>>> postgis-users at postgis.refractions.net
>>>>>>> http://postgis.refractions.net/mailman/listinfo/postgis-users
>>>>>> _______________________________________________
>>>>>> postgis-users mailing list
>>>>>> postgis-users at postgis.refractions.net
>>>>>> http://postgis.refractions.net/mailman/listinfo/postgis-users
>>>>> -- 
>>>>> Stephen Crawford
>>>>> Center for Environmental Informatics
>>>>> The Pennsylvania State University
>>>>> src176 at psu.edu
>>>>> 814.865.9905
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> postgis-users mailing list
>>>>> postgis-users at postgis.refractions.net
>>>>> http://postgis.refractions.net/mailman/listinfo/postgis-users
>>>> _______________________________________________
>>>> postgis-users mailing list
>>>> postgis-users at postgis.refractions.net
>>>> http://postgis.refractions.net/mailman/listinfo/postgis-users
> 

-- 
Bborie Park
Programmer
Center for Vectorborne Diseases
UC Davis
530-752-8380
bkpark at ucdavis.edu