[Zope-dev] Fault tolerance and thread-loss

[Tres Seaver]

"Chris McDonough" <chrism@digicool.com> wrote:
> 
> No.  You shouldn't be leaking threads.  Can you trace the thread
> leakage down to a single Product?
> 
> ----- Original Message -----
> From: "Jay, Dylan" <djay@avaya.com>
> To: <zope-dev@zope.org>
> Sent: Monday, December 18, 2000 7:00 PM
> Subject: [Zope-dev] Fault tolerance and thread-loss
> 
> 
> > I'm finding that my installation is blocking all requests after
> > awhile. By best guess would be that threads are being used and
> > somehow becoming permanently blocked. When all 4 are gone then
> > the server is useless and has to be restarted.
> >
> > Seems to me this is a real problem for fault tolerance. Perhaps
> > a monitor should be run on a separate thread that starts aborting
> > transactions when they have been running too long and there are
> > few remaining threads.
> >
> > Does this exist already?

I'm not really replying to either question, but guessing at the
underlying cause of the problem:  this smells *exactly* like the
behavior induced by the "multiple PUT/POST" race condition bug,
corrected by Evan Simpson's patched ZServer/HTTPServer.py:

  http://www.zope.org/Members/4am/postfixbug

Tres.
-- 
===============================================================
Tres Seaver                                tseaver@digicool.com
Digital Creations     "Zope Dealers"       http://www.zope.org