I was able to get consistent Zope failures on Solaris w/ DCOracle2 by disconnecting and reconnecting the connection in the Product management interface. If I run the server without the zdaemon, it prints that there was an interrupted system call on a "select()". Dario, this is Zope 2.4.3 running with Python 2.1.1 on a Solaris 2.8 server, with Oracle 8.1.7 and the latest recently released DCOracle2 driver. This is the first I've heard about Python 2.1 and Solaris not getting along. Is 2.2 any better?
On Thursday 13 December 2001 8:23 am, Jin Choi wrote:
I was able to get consistent Zope failures on Solaris w/ DCOracle2 by disconnecting and reconnecting the connection in the Product management interface. If I run the server without the zdaemon, it prints that there was an interrupted system call on a "select()".
Dario, this is Zope 2.4.3 running with Python 2.1.1 on a Solaris 2.8 server, with Oracle 8.1.7 and the latest recently released DCOracle2 driver.
This is the first I've heard about Python 2.1 and Solaris not getting along. Is 2.2 any better?
There is this which is fixed in both 2.2 and 2.1.2: http://sourceforge.net/tracker/?func=detail&atid=105470&aid=471942&group_id=... Is that the behaviour you're experiencing? Richard
It's not segfaulting. Running the start script with "-Z ''", it gives me: Traceback (most recent call last): File "/export/home/local/Zope-2.4.3-solaris-2.6-sparc/z2.py", line 774, in ? asyncore.loop() File "/usr/local/stow/Python-2.1.1/lib/python2.1/asyncore.py", line 194, in l\ oop poll_fun (timeout, map) File "/usr/local/stow/Python-2.1.1/lib/python2.1/asyncore.py", line 86, in po\ ll r,w,e = select.select (r,w,e, timeout) select.error: (4, 'Interrupted system call') I can trigger this by closing and attempting to reopen my Oracle connection. On Wednesday, December 12, 2001, at 05:17 PM, Richard Jones wrote:
On Thursday 13 December 2001 8:23 am, Jin Choi wrote:
I was able to get consistent Zope failures on Solaris w/ DCOracle2 by disconnecting and reconnecting the connection in the Product management interface. If I run the server without the zdaemon, it prints that there was an interrupted system call on a "select()".
Dario, this is Zope 2.4.3 running with Python 2.1.1 on a Solaris 2.8 server, with Oracle 8.1.7 and the latest recently released DCOracle2 driver.
This is the first I've heard about Python 2.1 and Solaris not getting along. Is 2.2 any better?
There is this which is fixed in both 2.2 and 2.1.2:
http://sourceforge.net/tracker/?func=detail&atid=105470&aid=471942&group_id= 5470
Is that the behaviour you're experiencing?
Richard
On Thursday 13 December 2001 9:13 am, Jin Choi wrote:
It's not segfaulting. Running the start script with "-Z ''", it gives me: Traceback (most recent call last): File "/export/home/local/Zope-2.4.3-solaris-2.6-sparc/z2.py", line 774, in ? asyncore.loop() File "/usr/local/stow/Python-2.1.1/lib/python2.1/asyncore.py", line 194, in l\ oop poll_fun (timeout, map) File "/usr/local/stow/Python-2.1.1/lib/python2.1/asyncore.py", line 86, in po\ ll r,w,e = select.select (r,w,e, timeout) select.error: (4, 'Interrupted system call')
I can trigger this by closing and attempting to reopen my Oracle connection.
Hrm - I've seen something about an asyncore replacement floating around one of these zope lists... maybe it's to blame... does anyone know anything about it? Richard
Problem solved. Thanks for your help. Looking through old mailing list archives, I see that this is a fairly well-known problem (but still not fixed in either asyncore.py or in Python!) Anyway, I took some code from ZEO's asyncwrap.py and modified asyncore. This really should get fixed once and for all.... *** asyncore.py~ Sun Sep 30 15:03:19 2001 --- asyncore.py Wed Dec 12 17:30:21 2001 *************** *** 50,55 **** --- 50,56 ---- import select import socket import sys + import errno import os if os.name == 'nt': *************** *** 83,89 **** r.append (fd) if obj.writable(): w.append (fd) ! r,w,e = select.select (r,w,e, timeout) if DEBUG: print r,w,e --- 84,97 ---- r.append (fd) if obj.writable(): w.append (fd) ! while 1: ! try: ! r,w,e = select.select (r,w,e, timeout) ! except select.error, err: ! if err[0] != errno.EINTR: ! raise ! else: ! break if DEBUG: print r,w,e On Wednesday, December 12, 2001, at 05:30 PM, Richard Jones wrote:
Hrm - I've seen something about an asyncore replacement floating around one of these zope lists... maybe it's to blame... does anyone know anything about it?
Richard
Problem solved. Thanks for your help. Looking through old mailing list archives, I see that this is a fairly well-known problem (but still not fixed in either asyncore.py or in Python!) Anyway, I took some code from ZEO's asyncwrap.py and modified asyncore. This really should get fixed once and for all....
FYI - the version of asyncore that will ship with Python 2.2 has the fixes for these things. In the meantime, for 2.5 I've added the 2.2 version to the distribution (+ some minor glue). It's not pretty, but it works :) Brian Lloyd brian@zope.com Software Engineer 540.361.1716 Zope Corporation http://www.zope.com
Matt Kromer sent a patch to asyncore to the Zope-dev list yesterday. This will fix this problem. ----- Original Message ----- From: "Richard Jones" <richard@bizarsoftware.com.au> To: "Jin Choi" <jsc@alum.mit.edu> Cc: <zope@zope.org> Sent: Wednesday, December 12, 2001 5:30 PM Subject: Re: [Zope] RE: Zope 2.4.3 crashing continually - asyncore replacement?
On Thursday 13 December 2001 9:13 am, Jin Choi wrote:
It's not segfaulting. Running the start script with "-Z ''", it gives me: Traceback (most recent call last): File "/export/home/local/Zope-2.4.3-solaris-2.6-sparc/z2.py", line 774, in ? asyncore.loop() File "/usr/local/stow/Python-2.1.1/lib/python2.1/asyncore.py", line 194, in l\ oop poll_fun (timeout, map) File "/usr/local/stow/Python-2.1.1/lib/python2.1/asyncore.py", line 86, in po\ ll r,w,e = select.select (r,w,e, timeout) select.error: (4, 'Interrupted system call')
I can trigger this by closing and attempting to reopen my Oracle connection.
Hrm - I've seen something about an asyncore replacement floating around one of these zope lists... maybe it's to blame... does anyone know anything about it?
Richard
_______________________________________________ Zope maillist - Zope@zope.org http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
We can't seem to get restarted after installing the patch. on 12/12/01 5:59 PM, Chris McDonough at chrism@zope.com wrote:
Matt Kromer sent a patch to asyncore to the Zope-dev list yesterday. This will fix this problem.
----- Original Message ----- From: "Richard Jones" <richard@bizarsoftware.com.au> To: "Jin Choi" <jsc@alum.mit.edu> Cc: <zope@zope.org> Sent: Wednesday, December 12, 2001 5:30 PM Subject: Re: [Zope] RE: Zope 2.4.3 crashing continually - asyncore replacement?
I think I've located the problem: http://lists.zope.org/pipermail/zope-dev/2001-August/012950.html It appears this problem has been known since at least October of last year! http://mail.python.org/pipermail/python-dev/2000-October/009672.html
participants (5)
-
Brian Lloyd -
Chris McDonough -
Chris Muldrow -
Jin Choi -
Richard Jones