in trunk

On Apr 5, 11:52 am, Gabriele  Alberti <gabriele.albe...@gmail.com>
wrote:
> Hello,
> I am not using connection pooling, and threaded=True is needed to
> avoid segfaults.
>
> I dont know why this happens, my wild guess is that cx_Oracle (or its
> underlying libs), when used in the same process, also if different
> threads, share something between different connections, needing
> locking..
>
> G.
>
> On Apr 5, 5:45 pm, mdipierro <mdipie...@cs.depaul.edu> wrote:
>
> > perhaps I misread your original post. Your patch suggested REMOVING
> > threaded=True and all I am saying is that it is not there. Are you
> > saying I misread your patch and you think threaded=True SHOULD be
> > there?
>
> > Are you using connection pooling? That may be why it is needed.
>
> > Massimo
>
> > On Apr 5, 6:23 am, Gabriele  Alberti <gabriele.albe...@gmail.com>
> > wrote:
>
> > > Hello,
> > > I am sorry but I have to disagree. I dont know exactly what happens as
> > > I am not into the DAL so much, but as I installed the latest version
> > > (running 1.76.5 now) with no threaded argument in the cx_Oracle
> > > connection I started getting back segfaults. They disappeared
> > > immediately after I put the threaded=True argument..
>
> > > If there's anything I can do to help, I'll be glad to.
>
> > > Regards,
>
> > > G.
>
> > > On Apr 5, 12:59 pm, Gabriele  Alberti <gabriele.albe...@gmail.com>
> > > wrote:
>
> > > > Hello,
> > > > yes I am running 1.76.1 (2010-03-01 19:56:55); I'll try the latest and
> > > > I'll let you know ASAP.
>
> > > > Thank you very much,
>
> > > > G.
>
> > > > On Apr 5, 5:43 am, mdipierro <mdipie...@cs.depaul.edu> wrote:
>
> > > > > Thank you Gabriele,
>
> > > > > is should definitively be threaded=False and not True since web2py
> > > > > does not share connection between threads. Yet, this is already the
> > > > > default. You must be running an older version of web2py.You should
> > > > > upgrade.
>
> > > > > Massimo
>
> > > > > On Apr 4, 11:33 am, Gabriele  Alberti <gabriele.albe...@gmail.com>
> > > > > wrote:
>
> > > > > > Hello,
> > > > > > I have been using web2py with oracle since few months now. With 
> > > > > > small
> > > > > > loads everything works fine, but when the load grows, often the 
> > > > > > python
> > > > > > process gets killed with a segfault; after few headaches to 
> > > > > > understand
> > > > > > what was going on, I think I spotted the problem and tried a 
> > > > > > solution
> > > > > > that works for me.
> > > > > > It seems the cx_Oracle libs when used in threaded programs (such as
> > > > > > the web2py server) needs to handle some locking, as suggested in the
> > > > > > cx_Oracle documentation found at
>
> > > > > >http://cx-oracle.sourceforge.net/html/module.html
>
> > > > > > [...]
> > > > > > The threaded argument is expected to be a boolean expression which
> > > > > > indicates whether or not Oracle should use the mode OCI_THREADED to
> > > > > > wrap accesses to connections with a mutex. Doing so in single 
> > > > > > threaded
> > > > > > applications imposes a performance penalty of about 10-15% which is
> > > > > > why the default is False.
> > > > > > [...]
>
> > > > > > I am not 100% sure this was the problem as I couldnt find a 
> > > > > > reasonable
> > > > > > way to prove it; however, since when I applied the following patch
> > > > > > (thats the only change I've made) I havent had crashes anymore.
> > > > > > So this is the simple patch I used and I propose
>
> > > > > > --- sql.py      2010-04-04 18:18:26.000000000 +0200
> > > > > > +++ sql_orig.py 2010-04-04 18:17:38.000000000 +0200
> > > > > > @@ -964,7 +964,7 @@
> > > > > >          elif self._uri[:9] == 'oracle://':
> > > > > >              self._dbname = 'oracle'
> > > > > >              self._pool_connection(lambda : \
> > > > > > -
> > > > > > cx_Oracle.connect(self._uri[9:],threaded=True))
> > > > > > +                                  cx_Oracle.connect(self._uri[9:]))
> > > > > >              self._cursor = self._connection.cursor()
> > > > > >              self._execute = lambda a: \
> > > > > >                  oracle_fix_execute(a,self._cursor.execute)
>
> > > > > > I'd like someone else to confirm this, and I really hope this can
> > > > > > help.
>
> > > > > > Kind regards,
>
> > > > > > G.

-- 
You received this message because you are subscribed to the Google Groups 
"web2py-users" group.
To post to this group, send email to web...@googlegroups.com.
To unsubscribe from this group, send email to 
web2py+unsubscr...@googlegroups.com.
For more options, visit this group at 
http://groups.google.com/group/web2py?hl=en.

Reply via email to