Starting Zope dies with an NT exception
I just copied across a working data.fs from my server to my development notebook, and now when I access localhost:8080, or even localhost:8080/manage, then python.exe dies with an NT exception. I copied the file again a couple of times to make sure it wasn't corrupted, but the same thing happens. And it's still working fine on my server. I also tried reinstalling Zope 2.1.6 on my notebook, but no dice. I presume that there is something about my data.fs that is corrupt, but should it kill python.exe ? Anything I can do to isolate the problem further? -- Regards, Graham Chiu gchiu<at>compkarori.co.nz http://www.compkarori.com/dynamo - The Homebuilt Dynamo http://www.compkarori.com/dbase - The dBase bulletin
Graham Chiu wrote:
I just copied across a working data.fs from my server to my development notebook, and now when I access localhost:8080, or even localhost:8080/manage, then python.exe dies with an NT exception.
What exception? Can you give us more detail than this?
I presume that there is something about my data.fs that is corrupt,
I doubt that.
but should it kill python.exe ? Anything I can do to isolate the problem further?
What is the error? -Michel
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 In article <38F3C679.D1C8E379@digicool.com>, Michel Pelletier <michel@digicool.com> writes
Graham Chiu wrote:
I just copied across a working data.fs from my server to my development notebook, and now when I access localhost:8080, or even localhost:8080/manage, then python.exe dies with an NT exception.
What exception? Can you give us more detail than this?
After starting up Zope, I can actually access a zwiki installation. When I point to localhost:8080, I get the authentication challenge. After entering my password, the frame separator appears on screen. Then Dr Watson appears saying that I have an application error in Python.exe. Waiting a bit longer, it says that there has been an Exception : access violation (0xc 0000005 ) at address 0x01651e67
I presume that there is something about my data.fs that is corrupt,
I doubt that.
I renamed my data.fs, and did a completely new Zope installation of 2.1.6 into the same directory as before. Zope now starts up ( though I see the tutorial has gone ). I stop the Zope service, copy back my old data.fs, and the problem recurs. I tried another data.fs that I had from 5 days ago that is running on another machine, and that also produces an Exception/access violation. I did not install any new software last night.
but should it kill python.exe ? Anything I can do to isolate the problem further?
What is the error?
-Michel
Suggestions? - -- Regards, Graham Chiu gchiu<at>compkarori.co.nz http://www.compkarori.com/dynamo - The Homebuilt Dynamo http://www.compkarori.com/dbase - The dBase bulletin -----BEGIN PGP SIGNATURE----- Version: PGPsdk version 1.7.1 iQA/AwUBOPNUYrTRdIWzaLpMEQK0zQCguZDq1VIJ6pqlzUguNq1EuiVzVtIAoPpi VxPi9pZHy3AsONWRO35aNDXx =P2BV -----END PGP SIGNATURE-----
On Wed, 12 Apr 2000 16:35:46 +1300, Graham Chiu <anon_emouse@hotmail.com> wrote:
I stop the Zope service, copy back my old data.fs, and the problem recurs.
Aha! you are running as a service. that makes a difference There is a bug in the win32 extensions where it is not correctly checking the parameters for CreateProcess. It lets some bad parameters though, and NT raises an exception (an NT exception, not a python exception) that eventually kills the process. I send a patch to Mark Hammond (who maintains this extension) - it's been included in his current version but not in the one shipped by DC. The fixable problem is that your registry is missing the command line that should be used to start the zope server. You can add this in by hand; it's documented in ZService.py Im sure I had posted a patch to the collector for this ages ago, but now I can't see it. Ill send another. Toby Dickenson tdickenson@geminidataloggers.com
In article <q5f8fssg1uibk171fagi9tjqmd475qhs2i@4ax.com>, Toby Dickenson <mbel44@dial.pipex.net> writes
I stop the Zope service, copy back my old data.fs, and the problem recurs.
Aha! you are running as a service. that makes a difference
Unfortunately that doesn't happen to be my problem. It crashes even when I run it as a batch file. I've tried a fresh installation. Okay. Then I copy my data.fs across without any of the optional products. Add Tiny tables. Okay. Then I add the ZODBC adapter, and wham, it dies. I recall now that I have seen, and reported this problem before. I have a search interface to my 30Mb database, and when users were entering single characters such as 'a', then the following query select * from mytable where comments like '%a%' also killed python.exe with the same error message. I'll have to do some more testing to see what is going on. -- Regards, Graham Chiu gchiu<at>compkarori.co.nz http://www.compkarori.com/dynamo - The Homebuilt Dynamo http://www.compkarori.com/dbase - The dBase bulletin
In article <lYoiaPATUE94Ew0Q@compkarori.com>, Graham Chiu <anon_emouse@hotmail.com> writes
I've tried a fresh installation. Okay. Then I copy my data.fs across without any of the optional products. Add Tiny tables. Okay. Then I add the ZODBC adapter, and wham, it dies.
Well, I think I have figured out what happened. When I copied my data.fs across, it had a ZODBC connection to a system DSN which on my notebook used a different driver but the same name. I ended up with two different ODBC drivers attaching to the same tables, and it seemed to be a lethal combination. I was using this before without a problem, but I then upgraded my sql engine in the weekend as well. In the course of this, I noticed another problem. Over the weekend I upgraded my server to 2.1.6 from running 2.1.3. This seemed to have killed a ZSQL method which was a straight forward insert. When tested from the management interface it worked fine. When tested in a document using literal values, it worked fine. When I used literal values within a method, it gave an ODBC error - about character or numeric truncation. I reverted to 2.1.3 and then noticed that the ZSQL method's icon had become a folder icon now, and I am unable to delete/access it. I recreated the ZSQL method, and it now works okay again. Questions: How can I remove a corrupt object like this? Is 2.1.6 absolutely solid on zsql/odbc? ( I realize that 2.1.5 was broken with respect to ZSQL methods ) Is it possible to keep earlier versions of Zope up on Zope.org? -- Regards, Graham Chiu gchiu<at>compkarori.co.nz http://www.compkarori.com/dynamo - The Homebuilt Dynamo http://www.compkarori.com/dbase - The dBase bulletin
participants (3)
-
Graham Chiu -
Michel Pelletier -
Toby Dickenson