Hi, we started a new service for our online newspaper, that is served by a Zope server. (http://rz-anzeigen.de) We currently experience the problem, that the machine serves really well and also during the high traffic times. But sometime in the early morning the Zope process crashes. Now I like to know if it is possible to get some more specific out from the Zope server when it crashes. Can I activate some kind of logging to monitor this, or something else that helps me to deal with this? Or is my only option to start the Server form the shell and see what traceback I get? Bye, Oliver
we started a new service for our online newspaper, that is served by a Zope server. (http://rz-anzeigen.de) We currently experience the problem, that the machine serves really well and also during the high traffic times. But sometime in the early morning the Zope process crashes. Now I like to know if it is possible to get some more specific out from the Zope server when it crashes. Can I activate some kind of logging to monitor this, or something else that helps me to deal with this? Or is my only option to start the Server form the shell and see what traceback I get?
I've been asking this list the very same question for quite a while. There are a few debug How-Tos that may help, but it's all very dependent on you're configuration as to how much extra work will be necessary. Can you give me more details of your configuration? What are you using for connection, are you going through Apache, if so using what service (PCGI,FCGI,HTTP_Proxy)? We have several sites up and running with Zope and have found that when there is /manage activity occuring during a time of moderate traffic we have frequent Zope restarts, it can't stay up more than 5-20 minutes. To date I've found no reliable method of tracking what is causing these, though I've created a bunch of monitoring and debuging tools to try. These tools are not written up yet, but I'd be happy to send out an explaination if you would like. It is also very easy to create code that will crash or loop the Zope server, these are easer to find since usually as soon as the object is called Zope reacts adversly, crashes, slows, etc. The problem using logs for this is log entries are usually written post completion. Good luck.
Tony Rossignol wrote:
we started a new service for our online newspaper, that is served by a Zope server. (http://rz-anzeigen.de) We currently experience the problem, that the machine serves really well and also during the high traffic times. But sometime in the early morning the Zope process crashes. Now I like to know if it is possible to get some more specific out from the Zope server when it crashes. Can I activate some kind of logging to monitor this, or something else that helps me to deal with this? Or is my only option to start the Server form the shell and see what traceback I get?
I've been asking this list the very same question for quite a while. There are a few debug How-Tos that may help, but it's all very dependent on you're configuration as to how much extra work will be necessary. Can you give me more details of your configuration? What are you using for connection, are you going through Apache, if so using what service (PCGI,FCGI,HTTP_Proxy)?
We have several sites up and running with Zope and have found that when there is /manage activity occuring during a time of moderate traffic we have frequent Zope restarts, it can't stay up more than 5-20 minutes. To date I've found no reliable method of tracking what is causing these, though I've created a bunch of monitoring and debuging tools to try. These tools are not written up yet, but I'd be happy to send out an explaination if you would like.
We've never seen behavior like this, can you elaborate?
It is also very easy to create code that will crash or loop the Zope server, these are easer to find since usually as soon as the object is called Zope reacts adversly, crashes, slows, etc. The problem using logs for this is log entries are usually written post completion.
This is kind of a distressing message Tony, we'd very much like to see any code that can easily crash Zope. -Michel
In article <38CC4BEA.730EF06A@digicool.com>, Michel Pelletier <michel@digicool.com> writes
This is kind of a distressing message Tony, we'd very much like to see any code that can easily crash Zope.
My experience is that returning a large dataset ( about 40Mb ) from a sql query will crash Zope. ------- Regards, Graham Chiu gchiu<at>compkarori.co.nz http://www.compkarori.com/dynamo - The Homebuilt Dynamo http://www.compkarori.com/dbase - The dBase bulletin
My experience is that returning a large dataset ( about 40Mb ) from a sql query will crash Zope.
Is there anyone else who has been having Zope crashes where SQL in related. We are using MySQL heavily in our applications. ZClasses are also being used.
On Tue, 14 Mar 2000, Graham Chiu wrote:
My experience is that returning a large dataset ( about 40Mb ) from a sql query will crash Zope.
I don't think zope should crash if this happens -- but is there any time when you'd be retrieving data that it wouldn't crash (insert name of your favorite web browser here, it doesn't matter which one)? ______ __ __ _____ _ _ | ____ | \_/ |_____] |_____| |_____| |_____ | | | | @ t w i s t e d m a t r i x . c o m http://www.twistedmatrix.com/~glyph/
In article <Pine.LNX.4.21.0003140032170.29324-100000@helix.twistedmatrix .com>, Glyph Lefkowitz <glyph@twistedmatrix.com> writes
My experience is that returning a large dataset ( about 40Mb ) from a sql query will crash Zope.
I don't think zope should crash if this happens -- but is there any time when you'd be retrieving data that it wouldn't crash (insert name of your favorite web browser here, it doesn't matter which one)?
This is a Zope crash, not a browser. Zope dies with a NT exception. 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:
In article <38CC4BEA.730EF06A@digicool.com>, Michel Pelletier <michel@digicool.com> writes
This is kind of a distressing message Tony, we'd very much like to see any code that can easily crash Zope.
My experience is that returning a large dataset ( about 40Mb ) from a sql query will crash Zope.
This could very well be a memory error. It would be interesting to know if the underlying libraries could handle a dataset this large, in this case it would not be Zope but the libraries (I can think of no reason why python couldn't handle a 40MB object, but I can immagine a poorly written python extension not doing this right, the old ct_sybase module is a perfect example). You didn't specify what database you are using. -Michel
In article <38CE748B.C562861F@digicool.com>, Michel Pelletier <michel@digicool.com> writes
written python extension not doing this right, the old ct_sybase module is a perfect example). You didn't specify what database you are using.
Interbase 5.6 with the ODBC adapter. ------- Regards, Graham Chiu gchiu<at>compkarori.co.nz http://www.compkarori.com/dynamo - The Homebuilt Dynamo http://www.compkarori.com/dbase - The dBase bulletin
On Sun, 12 Mar 2000, Oliver Andrich wrote:
crashes. Can I activate some kind of logging to monitor this, or something else that helps me to deal with this? Or is my only option to start the Server form the shell and see what traceback I get?
Do you know about STUPID_LOG_FILE? Put that variable in your environment when you start Zope, and Zope will write some stuff to the specified file. This info may or may not show something when Zope crashes, but it is worth a try. --RDM
Oliver Andrich wrote:
Hi,
we started a new service for our online newspaper, that is served by a Zope server. (http://rz-anzeigen.de) We currently experience the problem, that the machine serves really well and also during the high traffic times. But sometime in the early morning the Zope process crashes. Now I like to know if it is possible to get some more specific out from the Zope server when it crashes. Can I activate some kind of logging to monitor this, or something else that helps me to deal with this? Or is my only option to start the Server form the shell and see what traceback I get?
Try the DebugLog, this is an as yet undocumented feature of Zope: http://www.zope.org/Members/mcdonc/HowTos/DEBUG-LOG This will show various state transitions of requests. This helps pinpoint, for example, if your problem is a ZServer or an application issue. -Michel
participants (6)
-
Glyph Lefkowitz -
Graham Chiu -
Michel Pelletier -
Oliver Andrich -
R. David Murray -
Tony Rossignol