On 03/10/2008 1:38:36 PM +0100, Tobias Wiersch <[EMAIL PROTECTED]> wrote:
> Hi Gabriel,
> Gabriel Barazer schrieb:
>> I suppose this command is an init script and really does "apachectl -k 
>> graceful" ? Can you confirm this ? Since these init scripts are not part 
>> of the apache distribution, we don't know *exactly* what they do and 
>> there _is_ a big difference between the "-k graceful" and "-k restart" 
>> command for example.
> I checked it: the debian script will do a "graceful" restart.
>> then: strace -ff -r -p <PID>.
> Uh, sorry, my last strace was without -ff ... I'm not familiar with strace.
> Here it comes again:
> =========================
> Traced process 1:
>      0.000000 poll([{fd=15, events=POLLIN, revents=POLLIN}], 1, 3000) = 1
>      2.180921 read(15, "/home/xxx.de/htdocs/testxxx.php"..., 12840) = 12840

OK, this means the fcgid PM is sending 12840 bytes over the socket (fd 
15 in your case is probably the fastcgi socket listened).

1) Is the delay *exactly* 2 seconds when calling a PHP script (before 
getting the http error 500 from your suexec test) ? Or is there another 
additionnal delay to look for ?

2) have you tried to do the same using a localhost connection? e.g using 
time ab -c 1 -n 1

Because fcgid has a major design flaw by buffering the data between 
apache and the fastcgi process, the delay could come from this buffering 
step. You don't have any network delay when sending data over localhost 
so if the problem is related to buffering, this test would confirm it.

Oh, and add something like -s 4096 to the strace command to see more on 
the socket read.


This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
Mod-fcgid-users mailing list

Reply via email to