I'm using fastcgipp to handle multi request in my application. But when the concurrent request number up to 2000 - 3000 in one process, the fastcgipp can not handle new request from nginx(in nginx the error:connect to ***.socket faild(11 resource temporaly unavailable)).
At this time when I use strace to watch the fastcgipp application, the message like below:
20:23:14.588074 poll({fdlist} {fdlist) <0.001073>
20:23:14.589414 accept(0, {domain socket path}, [16]) = 57360 <0.000061>
20:23:14.589536 fcntl(57360, F_GETFL) = 0x2 (flags O_RDWR) <0.000019>
20:23:14.589597 fcntl(57360, F_SETFL, O_RDWR) = 0 <0.000013>
20:23:14.589654 read(57360, "\1\1\0\1\0\10\0\0", 8) = 8 <0.000057>
20:23:14.589901 read(57360, "\0\1\0\0\0\0\0\0", 8) = 8 <0.000013>
In the applcation the behavior repeat up six lines. It seems like the application accept new connection, which can not read the request content.
when I close client, the application's cpu usage can not decrease, the strace message like below:
20:28:57.043388 poll([fdlist])
20:28:57.059699 poll([fdlist])
the behavior repeat execute poll.
I'm confused why the request can not pass to the response function to handle?