[mod_python] Multiple interpreter instances?

Stefan Hudac stevohudac at yahoo.com
Wed Jul 14 21:24:13 EDT 2004


Hi guys,

Thanks for your great work. Thanks to folks like you our work is more 
fun. Anyway I have a question to ask.

I am trying to create application framework on top of Apache. From my 
tests it looks like Python interpreter is being spawned per Apache fork. 
Practically meaning if there is any long-time-running process created 
through mod_python it is actually in memory as many times as many forks 
there are.

I found out by having an error in scripts. I did my tests resulting in 
an error. When I fixed a problem, I redownloaded a page. Looks like 
original fork was killed by Apache or another fork answered my request 
this time. It took a while to launch the framework and error was 
corrected. I reloaded the page and just corrected problem popped on me 
again... I was able to repeat this behavior couple times by 
redownloading a page. Everything got resolved when I restarted Apache.

I'd like to know if my observations are correct. If yes it means all 
resources kept in memory are actually unaccessible among Apache forks, 
therefore multiplicated as many times as there are forks. In the worst 
scenario it may mean that data created by objects executed through 
mod_python and kept in memory are not consistent among different Apache 
forks. Am I right?

stefan



More information about the Mod_python mailing list