Opened 3 years ago
Closed 3 years ago
#5023 closed defect (fixed)
berrie64 and berrie hanging on garden test "ST_MaximumInscribedCircle with malformed geometries"
Reported by: | robe | Owned by: | robe |
---|---|---|---|
Priority: | blocker | Milestone: | PostGIS GEOS |
Component: | QA/buildbots | Version: | 3.1.x |
Keywords: | Cc: |
Description (last modified by )
berrie64 got swapped out with a newer berrie64 and sadly she is hanging on her garden test run.
00:18:40 PostgreSQL 13.5 on aarch64-unknown-linux-gnu, compiled by gcc (Debian 10.2.1-6) 10.2.1 20210110, 64-bit 00:18:40 Postgis 3.1.5dev - (77a4e86) - 2021-11-22 05:04:52 00:18:40 scripts 3.1.5dev 77a4e86 00:18:40 raster scripts 3.1.5dev 77a4e86 00:18:40 GEOS: 3.9.0-CAPI-1.16.2 00:18:40 PROJ: 7.2.1 00:18:40 GDAL: GDAL 3.2.2, released 2021/03/05 00:18:40 00:26:22 ------------------------------------------------- 00:26:22 Regression tests in progress (it will take time) 00:26:22 Result output: ./regress/postgis_garden_result.txt 00:26:22 ------------------------------------------------- 00:26:22 psql --no-psqlrc -ad postgis_garden < ../doc/postgis_gardentest_31.sql > postgis_garden_result.txt 2>&1 17:07:08 make[1]: *** [Makefile:100: garden] Error 2 17:07:08 make[1]: Leaving directory '/home/jenkins/workspace/PostGIS_Worker_Run/label/berrie64/77a4e86e2a22432d6bd56b424d8a48517d91f317/regress' 17:07:08 make: *** [GNUmakefile:115: garden] Error 2 17:07:08 Build step 'Execute shell' marked build as failure
also on PostGIS 3.2 /PG 14
19:53:33 PostgreSQL 14.1 on aarch64-unknown-linux-gnu, compiled by gcc (Debian 10.2.1-6) 10.2.1 20210110, 64-bit 19:53:33 Postgis 3.2.0dev - (677d201) - 2021-11-21 00:40:23 19:53:33 scripts 3.2.0dev 677d201 19:53:33 raster scripts 3.2.0dev 677d201 19:53:33 GEOS: 3.9.0-CAPI-1.16.2 19:53:33 PROJ: 7.2.1 19:53:33 GDAL: GDAL 3.2.2, released 2021/03/05 20:00:26 ------------------------------------------------- 20:00:26 Regression tests in progress (it will take time) 20:00:26 Result output: ./regress/postgis_garden_result.txt 20:00:26 ------------------------------------------------- 20:00:26 psql --no-psqlrc -ad postgis_garden < ../doc/postgis_gardentest_32.sql > postgis_garden_result.txt 2>&1 22:21:32 make[1]: *** [Makefile:100: garden] Error 2 22:21:32 make[1]: Leaving directory '/home/jenkins/workspace/PostGIS_Worker_Run/label/berrie64/677d2019a7b7da0f4a6b63eac22dd607ffd72a83/regress' 22:21:32 make: *** [GNUmakefile:115: garden] Error 2 22:21:32 Build step 'Execute shell' marked build as failure 22:21:38 Finished: FAILURE
I'll update with forensics after I have checked the forensics
Change History (12)
comment:2 by , 3 years ago
Summary: | berrie64 failing on garden tests → berrie64 failing on garden tests ST_MaximumInscribedCircle with malformed geometries |
---|
comment:3 by , 3 years ago
okay not that -- just recompiled 3.2 with GEOS 3.9
I'll try 3.1 next
POSTGIS="3.2.0dev 3.2.0beta1-25-g540e03f0e" [EXTENSION] PGSQL="140" GEOS="3.9.0-CAPI-1.14.1" PROJ="7.2.1" GDAL="GDAL 3.3.3, released 2021/10/25" LIBXML="2.9.9" LIBJSON="0.12" LIBPROTOBUF="1.2.1" WAGYU="0.5.0 (Internal)" TOPOLOGY RASTER
No error.
comment:4 by , 3 years ago
okay just tried against 3.1 branch and no issues there either:
POSTGIS="3.1.5dev 3.1.4-17-g77a4e86e2" [EXTENSION] PGSQL="140" GEOS="3.9.0-CAPI-1.14.1" SFCGAL="1.3.10" PROJ="7.2.1" GDAL="GDAL 3.3.3, released 2021/10/25" LIBXML="2.9.9" LIBJSON="0.12" LIBPROTOBUF="1.2.1" WAGYU="0.5.0 (Internal)" TOPOLOGY RASTER
can't imagine it's a PG13 issue. I'll run thru the garden tests later.
comment:5 by , 3 years ago
Description: | modified (diff) |
---|
comment:7 by , 3 years ago
Same issue on berrie (32-bit Raspberry Pi arm). So my conclusion is it must be an ARM / GEOS >= 3.9+. I'm going to check against the GEOS built by berrie instead of using the system provided (3.9.0) and see if perhaps it's fixed in newer GEOS. and get a backtrace.
FWIW, it isn't failing. It's going into some sort of infinite loop.
comment:8 by , 3 years ago
Summary: | berrie64 failing on garden tests ST_MaximumInscribedCircle with malformed geometries → berrie64 and berrie hanging on garden test "ST_MaximumInscribedCircle with malformed geometries" |
---|
comment:9 by , 3 years ago
Sadly, seems to still hang with latest GEOS. I'm going to commit my changes soon to both geos and postgis, to have the berries build against latest head of GEOS so we can monitor this issue.
https://debbie.postgis.net/job/PostGIS_Worker_Run/2973/label=berrie/consoleFull
22:14:08 -------------- Dependencies -------------- 22:14:08 GEOS config: /home/jenkins/workspace/geos/rel-3.11.0/bin/geos-config 22:14:08 GEOS version: 3.11.0dev 22:14:08 GDAL config: /usr/bin/gdal-config 22:14:08 GDAL version: 3.2.2 22:14:08 PostgreSQL config: /home/jenkins/workspace/pg/label/berrie/rel/pg14w32/bin/pg_config 22:14:08 PostgreSQL version: PostgreSQL 14.1 22:14:08 PROJ4 version: 72 22:14:08 Libxml2 config: /usr/bin/xml2-config 22:14:08 Libxml2 version: 2.9.10 22:14:08 JSON-C support: yes 22:14:08 protobuf support: no 22:14:08 PCRE support: yes 22:14:08 Perl: /usr/bin/perl
comment:10 by , 3 years ago
"Good" news, I'm getting an infinite loop on MacOS, on GEOS main, I have it down to
select ST_MaximumInscribedCircle('POLYGON((1 2,1 2,1 2,1 2,3 2,1 2))');
Going to pull it back into a GEOS unit test now.
comment:12 by , 3 years ago
Milestone: | PostGIS 3.1.5 → PostGIS GEOS |
---|---|
Resolution: | → fixed |
Status: | new → closed |
I'm going to close this out and will trigger berrie64 to do gardens as well if berrie stops hanging after building new geos.
okay the offensive query is this one:
Works fine on my system:
, so could be a GEOS 3.9.0 bug fixed in later GEOS.