Re: apache style squid?

From: Oskar Pearson <oskar@dont-contact.us>
Date: Tue, 7 Oct 1997 20:17:05 +0200

--MimeMultipartBoundary
Content-Type: text/plain; charset=us-ascii

Hi

Just some stats - everyone, please DONT post millions of these
messages to the list - I am doing this since I haven't received any
others.

> > linux do 100 fork()s and exit()s per second, while doing other things. An
This was definitely doing other things :)

> Small test on my PPro 200 Solaris 2.5.1:

Running the same test on a linux-2.0.31pre10 Pentium (classic) 133 (64M) I get
this: (It's completely idle)

transproxy:~ # strace -c ./forktest
execve("./forktest", ["./forktest"], [/* 14 vars */]) = 0
% time seconds usecs/call calls errors syscall
------ ----------- ----------- --------- --------- ----------------
 96.63 0.638297 638 1000 fork
  3.18 0.021020 21 1000 wait4
  0.05 0.000320 53 6 mmap
  0.04 0.000243 122 2 open
  0.03 0.000219 219 1 read
  0.02 0.000165 28 6 mprotect
  0.02 0.000153 77 2 1 stat
  0.01 0.000054 54 1 munmap
  0.00 0.000028 14 2 close
  0.00 0.000021 21 1 personality
  0.00 0.000014 14 1 getegid
  0.00 0.000012 12 1 getgid
  0.00 0.000010 10 1 getuid
  0.00 0.000010 10 1 geteuid
------ ----------- ----------- --------- --------- ----------------
100.00 0.660566 2025 1 total

> sys totals: .94 2023 2
> usr time: .01
> elapsed: 4.34
> truss accounts only for parent calls, but just /bin/time shows that all 1000 forks/exits/waits
> are done in 4 secs. This makes roughly 250 forks-exits-waits per second.

transproxy:~ # time ./forktest

real 0m1.223s
user 0m0.080s
sys 0m1.150s

That's about 800 on a machine a LOT slower...

Oskar

--MimeMultipartBoundary--
Received on Tue Jul 29 2003 - 13:15:43 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:11:26 MST