[web2py] Re: potential memory leak???

2011-03-04 Thread Anthony
On Friday, March 4, 2011 1:41:13 AM UTC-5, VP wrote: Apache configuration: 5 processes, 1 thread (more than 1 threads will cause WSGI premature script error as previously described). I thought the WSGI premature script error problem was resolved by upgrading psycopg2 to a version

[web2py] Re: potential memory leak???

2011-03-04 Thread Massimo Di Pierro
I do not think it is a memory leak. Please run two tests: 1) What do you get with run with the built-in server instead of apache? 2) what is the output of print db(db.data).select('sum(length(data.body) +length(data.body1)+length(data.body2)+length(data.body3))') I suspect it may be a big

[web2py] Re: potential memory leak???

2011-03-04 Thread VP
Massimo: 1. This is with Apache. 2. print db(db.data).select('sum(length(data.body) +length(data.body1)+length(data.body2)+length(data.body3))') Answer: 1764948 I am looking at ps aux right now. Under VSZ column, there's 141M. Resource is still not released since last night. On Mar 4, 8:27 am,

[web2py] Re: potential memory leak???

2011-03-04 Thread VP
For some reason, I missed the rest of that discussion. I didn't upgrade psycopg2. My VPS has not upgraded to Debian 6 yet (but will soon). So, I will wait for a few weeks and then this module will be upgraded together with Debian 6. I will report on this later. On Mar 4, 8:15 am, Anthony

[web2py] Re: potential memory leak???

2011-03-04 Thread ron_m
RSS (Resident Set Size) is a closer approximation of actual physical memory consumed by a process. VSZ includes shared libraries and anything else required to be mapped in to make the process complete but shared libraries are not in physical memory more than once for all processes except the

[web2py] Re: potential memory leak???

2011-03-04 Thread VP
RSS is smaller than VSZ, but the behavior appeared to be the same. The point is that accumulated resources do not seem to get released. I have made changes (upgraded psycopg2), so I will report on this with further testing. On Mar 4, 11:15 am, ron_m ron.mco...@gmail.com wrote: RSS (Resident

[web2py] Re: potential memory leak???

2011-03-04 Thread ron_m
Depending on the system type the memory doesn't get released. It is rare that a malloc implementation calls sbreak with a smaller number than the previous call which is what you are looking for. If you have a program that consumes say 100 MB to do some initialization and then frees all those