[postgis-users] Signal 11: Segmentation fault

[Paul Ramsey]

Actually, this doesn't match our fixed but, which was an index
node-splitting bug that only triggered when new data was inserted. But
your crash is happening then you query. The difference between the
whole table and the limit query points to the problem being a specific
record. The fact that you could dump the table and restore it and
*still* get the problem is very surprising. The idea that it could all
work for a while and then simply start failing is the most confusing
part of all.

Can you share

(a) a dump of your table
(b) the queries you run
(c) the exact versions you are running
(d) your operating system information
(e) your hardware platform

If we can replicate the bug, we can probably fix it. You don't have to
share your info publicly, just email me and/or strk.

P.

On Fri, Jul 20, 2012 at 12:49 AM, Rebecca Clarke <r.clarke83 at gmail.com> wrote:
> But that doesn't explain why it's just started happening. Can you give me a
> link that tells me what the bug is.
> Cheers
>
>
> On Thu, Jul 19, 2012 at 6:22 PM, Sandro Santilli <strk at keybit.net> wrote:
>>
>> Sounds like a bad bug recently discovered and already fixed in the
>> repository. You'll have to upgrade to 1.5.5, but it isn't out yet.
>>
>> If you filed a bug it may put some more pressure to the release manager :)
>>
>> --strk;
>>
>> On Thu, Jul 19, 2012 at 05:37:12PM +0100, Rebecca Clarke wrote:
>> > Hi there
>> >
>> > I'm running Postgresql 8.4 with postgis 1.5.
>> > I have a table that contains roughly 2000 records.
>> > The geometry in the table is a polygon.
>> >
>> > I can run the following query with no problems
>> >
>> > select the_geom from mytable;
>> >
>> > However when I run something like this:
>> >
>> > select st_centroid(the_geom) from mytable;
>> >
>> > or
>> >
>> > select st_isvalid(the_geom) from mytable;
>> >
>> > I get the following error everytime:
>> >
>> > 2012-07-19 17:23:47 BST LOG:  server process (PID 19445) was terminated
>> > by
>> > signal 11: Segmentation fault
>> > 2012-07-19 17:23:47 BST LOG:  terminating any other active server
>> > processes
>> > 2012-07-19 17:23:47 BST LOG:  all server processes terminated;
>> > reinitializing
>> > 2012-07-19 17:23:47 BST LOG:  database system was interrupted; last
>> > known
>> > up at 2012-07-19 17:23:33 BST
>> > 2012-07-19 17:23:47 BST LOG:  database system was not properly shut
>> > down;
>> > automatic recovery in progress
>> > 2012-07-19 17:23:47 BST LOG:  record with zero length at 1406/B6388B94
>> > 2012-07-19 17:23:47 BST LOG:  redo is not required
>> > 2012-07-19 17:23:47 BST LOG:  autovacuum launcher started
>> > 2012-07-19 17:23:47 BST LOG:  database system is ready to accept
>> > connections
>> >
>> >
>> > If I limit the query to 100 records it works.
>> >
>> > In the same table I have another geometry that is a point. When I run
>> > the
>> > st_isvalid against that, it does not error.
>> >
>> > What is strange is this has only just started happening. I thought it
>> > was a
>> > hardware issue, so I completely dumped the database and recreated it on
>> > another server but the issue persisted. I'm not sure what I have done
>> > wrong.
>> >
>> >
>> > The data originates from an mdb. The geom's were stored as OLE Objects.
>> > When I imported them into postgres they were bytea. I converted them
>> > with
>> > st_geomfromewkb(the_geom).
>> >
>> > Any help or directions on where I can go to figure out how to solve this
>> > issue would be great.
>> >
>> > Thanks
>> >
>> > Rebecca
>> _______________________________________________
>> 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
>