Re: visit_decref: Assertion `gc->gc.gc_refs != 0' failed.

Michael Tharp <gxti-1pawZKhx9Om5WRpDikjj11aTQe2KTcn/@public.gmane.org> Mon, 29 Mar 2010 11:30:36 -0400
Newsgroups gmane.comp.python.db.psycopg.devel
Message-ID <[email protected]>
On 03/26/2010 10:20 PM, Jan Urbański wrote:
> On 27/03/10 03:17, Jan Urbański wrote:
>> Here's a patch that fixes my problem. The issue seemed to be that
>> psycopg2 was storing the asynchronous cursor in conn->async_cursor, and
>
> Dang, previous patch had the actual reffing commented, sorry :| Correct
> patch attached.
>
> Jan

Probably not related. I think my crash has something to do with the type 
registration system because A) the object crashing it is always a 
psycopg2._psycopg.type and B) the object that was being allocated at the 
time is always my type. Of particular note is that my type subclasses 
long, so maybe it only happens for sub-classes of built-ins.

Unfortunately, I can't distill this to a simple test case because it's 
non-deterministically triggered during a load test. I tried breaking it 
down to just selecting the type in a tight loop and nothing happened 
after hours of run time.

I will take another few passes over the type code and see if maybe 
there's a problem similar to the one addressed in your patch. Thanks for 
the suggestion, though!

-- m. tharp
_______________________________________________
Psycopg mailing list
[email protected]
http://lists.initd.org/mailman/listinfo/psycopg