Hi guys,
I am trying to find out what is causing lockf problem on php-fpm web servers (using lighttpd + xcache - opcode cache on). I try to truss some PIDs and found out this:
of course there are many php to include, and if all of them taking 0.07 seconds to stat(), lets say with just 10 includes we are already on 0.7 seconds. I think thats too much...
Any idea whats going on ?
Those files are just includes in php.
I am trying to find out what is causing lockf problem on php-fpm web servers (using lighttpd + xcache - opcode cache on). I try to truss some PIDs and found out this:
Code:
0.606608573 open("/data/www/public_html/libs/xslt-php4-to-php5.php",O_RDONLY,00) = 6 (0x6)
0.606860279 stat("/data/www/public_html/libs/xslt-php4-to-php5.php",{ mode=-rwxr-xr-x ,inode=22801636,size=3856,blksize=16384 }) = 0 (0x0)
0.670303811 fcntl(3,F_SETLKW,0x7fffffffa370) *beep* *beep* = 0 (0x0)of course there are many php to include, and if all of them taking 0.07 seconds to stat(), lets say with just 10 includes we are already on 0.7 seconds. I think thats too much...
Code:
mount: /dev/mirror/gm0s1f on / (ufs, local, noatime, soft-updates)Those files are just includes in php.