Skip to content

Instantly share code, notes, and snippets.

@shogo82148

shogo82148/app.psgi

Last active October 12, 2015 11:40
Show Gist options
  • Save shogo82148/5150dccc7cbe1b33b5fa to your computer and use it in GitHub Desktop.
Save shogo82148/5150dccc7cbe1b33b5fa to your computer and use it in GitHub Desktop.
Download ZIP
testing Starlet/Gazelle graceful shutdown
Raw
app.psgi
# start_server --port 5000 --pid app.pid -- plackup -s Starlet --max-workers 20 app.psgi > /dev/null 2>&1
# while true; do kill -HUP `cat app.pid`; sleep 1; done
my $app = sub {
my $env = shift;
return [
200,
[ 'Content-Type' => 'text/plain' ],
[ "Hello World" ],
];
};
Raw
do-not-send-kill-hup-Gazelle
$ ab -c 20 -n 100000 -r http://localhost:5000/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 10000 requests
Completed 20000 requests
Completed 30000 requests
Completed 40000 requests
Completed 50000 requests
Completed 60000 requests
Completed 70000 requests
Completed 80000 requests
Completed 90000 requests
Completed 100000 requests
Finished 100000 requests
Server Software: gazelle
Server Hostname: localhost
Server Port: 5000
Document Path: /
Document Length: 11 bytes
Concurrency Level: 20
Time taken for tests: 10.294 seconds
Complete requests: 100000
Failed requests: 0
Write errors: 0
Total transferred: 12900774 bytes
HTML transferred: 1100066 bytes
Requests per second: 9714.85 [#/sec] (mean)
Time per request: 2.059 [ms] (mean)
Time per request: 0.103 [ms] (mean, across all concurrent requests)
Transfer rate: 1223.92 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 1 0.4 0 10
Processing: 0 1 0.7 2 36
Waiting: 0 1 0.7 1 36
Total: 0 2 0.7 2 36
ERROR: The median and mean for the initial connection time are more than twice the standard
deviation apart. These results are NOT reliable.
WARNING: The median and mean for the processing time are not within a normal deviation
These results are probably not that reliable.
Percentage of the requests served within a certain time (ms)
50% 2
66% 2
75% 2
80% 2
90% 3
95% 3
98% 3
99% 4
100% 36 (longest request)
Raw
do-not-send-kill-hup-Starlet
$ ab -c 20 -n 100000 -r http://localhost:5000/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 10000 requests
Completed 20000 requests
Completed 30000 requests
Completed 40000 requests
Completed 50000 requests
Completed 60000 requests
Completed 70000 requests
Completed 80000 requests
Completed 90000 requests
Completed 100000 requests
Finished 100000 requests
Server Software: Plack::Handler::Starlet
Server Hostname: localhost
Server Port: 5000
Document Path: /
Document Length: 11 bytes
Concurrency Level: 20
Time taken for tests: 20.319 seconds
Complete requests: 100000
Failed requests: 0
Write errors: 0
Total transferred: 14500725 bytes
HTML transferred: 1100055 bytes
Requests per second: 4921.53 [#/sec] (mean)
Time per request: 4.064 [ms] (mean)
Time per request: 0.203 [ms] (mean, across all concurrent requests)
Transfer rate: 696.93 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 1 1.0 1 13
Processing: 0 3 2.5 3 104
Waiting: 0 3 2.5 3 104
Total: 1 4 2.5 4 105
Percentage of the requests served within a certain time (ms)
50% 4
66% 4
75% 4
80% 5
90% 6
95% 6
98% 8
99% 9
100% 105 (longest request)
Raw
my-patch-Gazelle
$ ab -c 20 -n 100000 -r http://localhost:5000/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 10000 requests
Completed 20000 requests
Completed 30000 requests
Completed 40000 requests
Completed 50000 requests
Completed 60000 requests
Completed 70000 requests
Completed 80000 requests
Completed 90000 requests
Completed 100000 requests
Finished 100000 requests
Server Software: gazelle
Server Hostname: localhost
Server Port: 5000
Document Path: /
Document Length: 11 bytes
Concurrency Level: 20
Time taken for tests: 11.251 seconds
Complete requests: 100000
Failed requests: 0
Write errors: 0
Total transferred: 12901548 bytes
HTML transferred: 1100132 bytes
Requests per second: 8888.15 [#/sec] (mean)
Time per request: 2.250 [ms] (mean)
Time per request: 0.113 [ms] (mean, across all concurrent requests)
Transfer rate: 1119.83 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 1 0.7 1 20
Processing: 0 1 0.7 2 27
Waiting: 0 1 0.7 1 27
Total: 0 2 1.0 2 28
WARNING: The median and mean for the processing time are not within a normal deviation
These results are probably not that reliable.
Percentage of the requests served within a certain time (ms)
50% 2
66% 2
75% 2
80% 3
90% 3
95% 3
98% 4
99% 5
100% 28 (longest request)
Raw
my-patch-Starlet
$ ab -c 20 -n 100000 -r http://localhost:5000/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 10000 requests
Completed 20000 requests
Completed 30000 requests
Completed 40000 requests
Completed 50000 requests
Completed 60000 requests
Completed 70000 requests
Completed 80000 requests
Completed 90000 requests
Completed 100000 requests
Finished 100000 requests
Server Software: Plack::Handler::Starlet
Server Hostname: localhost
Server Port: 5000
Document Path: /
Document Length: 11 bytes
Concurrency Level: 20
Time taken for tests: 22.405 seconds
Complete requests: 100000
Failed requests: 0
Write errors: 0
Total transferred: 14500725 bytes
HTML transferred: 1100055 bytes
Requests per second: 4463.36 [#/sec] (mean)
Time per request: 4.481 [ms] (mean)
Time per request: 0.224 [ms] (mean, across all concurrent requests)
Transfer rate: 632.05 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 1 1.4 1 29
Processing: 0 3 1.9 3 147
Waiting: 0 3 1.9 3 147
Total: 1 4 2.2 4 149
Percentage of the requests served within a certain time (ms)
50% 4
66% 5
75% 5
80% 6
90% 6
95% 8
98% 9
99% 11
100% 149 (longest request)
Raw
patch-Starlet.diff
diff --git a/lib/Starlet/Server.pm b/lib/Starlet/Server.pm
index 520cd37..2c6afae 100644
--- a/lib/Starlet/Server.pm
+++ b/lib/Starlet/Server.pm
@@ -125,14 +125,10 @@ sub accept_loop {
my($self, $app, $max_reqs_per_child) = @_;
my $proc_req_count = 0;
- $self->{can_exit} = 1;
my $is_keepalive = 0;
local $SIG{TERM} = sub {
- exit 0 if $self->{can_exit};
$self->{term_received}++;
- exit 0
- if ($is_keepalive && $self->{can_exit}) || $self->{term_received} > 1;
- # warn "server termination delayed while handling current HTTP request";
+ exit 0 if $self->{term_received} > 1;
};
local $SIG{PIPE} = 'IGNORE';
@@ -207,6 +203,9 @@ sub _get_acceptor {
if (my ($conn, $peer) = $listen->{sock}->accept) {
return ($conn, $peer, $listen);
}
+ elsif ($! == EINTR) {
+ exit 0 if $self->{term_received};
+ }
}
};
}
@@ -228,10 +227,21 @@ sub _get_acceptor {
return sub {
while (1) {
while (! flock($lock_fh, LOCK_EX)) {
- next if $! == EINTR;
+ if ($! == EINTR) {
+ exit 0 if $self->{term_received};
+ next;
+ }
die "failed to lock file:@{[$self->{lock_path}]}:$!";
}
my $nfound = select(my $rout = $rin, undef, undef, undef);
+ if ($nfound == -1) { # trap err
+ if ($! == EINTR) {
+ flock($lock_fh, LOCK_UN);
+ exit 0 if $self->{term_received};
+ next;
+ }
+ die "failed to select file:@{[$self->{lock_path}]}:$!";
+ }
for (my $i = 0; $nfound > 0; ++$i) {
my $fd = $fds[$i];
next unless vec($rout, $fd, 1);
@@ -241,6 +251,9 @@ sub _get_acceptor {
flock($lock_fh, LOCK_UN);
return ($conn, $peer, $listen);
}
+ elsif ($! == EINTR) {
+ exit 0 if $self->{term_received};
+ }
}
flock($lock_fh, LOCK_UN);
}
@@ -256,7 +269,6 @@ sub handle_connection {
my $pipelined_buf='';
my $res = $bad_response;
- local $self->{can_exit} = (defined $prebuf) ? 0 : 1;
while (1) {
my $rlen;
if ( $rlen = length $prebuf ) {
@@ -269,7 +281,6 @@ sub handle_connection {
$is_keepalive ? $self->{keepalive_timeout} : $self->{timeout},
) or return;
}
- $self->{can_exit} = 0;
my $reqlen = parse_http_request($buf, $env);
if ($reqlen >= 0) {
# handle request
Raw
send-kill-hup-Gazelle
$ ab -c 20 -n 100000 -r http://localhost:5000/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 10000 requests
Completed 20000 requests
Completed 30000 requests
Completed 40000 requests
Completed 50000 requests
Completed 60000 requests
Completed 70000 requests
Completed 80000 requests
Completed 90000 requests
Completed 100000 requests
Finished 100000 requests
Server Software: gazelle
Server Hostname: localhost
Server Port: 5000
Document Path: /
Document Length: 11 bytes
Concurrency Level: 20
Time taken for tests: 11.350 seconds
Complete requests: 100000
Failed requests: 13
(Connect: 0, Receive: 0, Length: 13, Exceptions: 0)
Write errors: 0
Total transferred: 12899226 bytes
HTML transferred: 1099934 bytes
Requests per second: 8810.82 [#/sec] (mean)
Time per request: 2.270 [ms] (mean)
Time per request: 0.113 [ms] (mean, across all concurrent requests)
Transfer rate: 1109.89 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 1 0.6 1 22
Processing: 0 2 0.8 2 28
Waiting: 0 1 0.8 1 28
Total: 0 2 1.0 2 29
Percentage of the requests served within a certain time (ms)
50% 2
66% 2
75% 2
80% 3
90% 3
95% 4
98% 4
99% 5
100% 29 (longest request)
Raw
send-kill-hup-Starlet
$ ab -c 20 -n 100000 -r http://localhost:5000/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 10000 requests
Completed 20000 requests
Completed 30000 requests
Completed 40000 requests
Completed 50000 requests
Completed 60000 requests
Completed 70000 requests
Completed 80000 requests
Completed 90000 requests
Completed 100000 requests
Finished 100000 requests
Server Software: Plack::Handler::Starlet
Server Hostname: localhost
Server Port: 5000
Document Path: /
Document Length: 11 bytes
Concurrency Level: 20
Time taken for tests: 22.619 seconds
Complete requests: 100000
Failed requests: 135
(Connect: 0, Receive: 45, Length: 45, Exceptions: 45)
Write errors: 0
Total transferred: 14495505 bytes
HTML transferred: 1099659 bytes
Requests per second: 4421.02 [#/sec] (mean)
Time per request: 4.524 [ms] (mean)
Time per request: 0.226 [ms] (mean, across all concurrent requests)
Transfer rate: 625.83 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 1 1.4 1 26
Processing: 0 3 2.0 3 81
Waiting: 0 3 2.0 3 80
Total: 1 5 2.2 4 81
Percentage of the requests served within a certain time (ms)
50% 4
66% 5
75% 5
80% 6
90% 7
95% 8
98% 10
99% 12
100% 81 (longest request)
Raw
send-kill-hup-Starman
$ start_server --port 5000 --pid app.pid -- plackup -s Starman --workers 20 app.psgi
$ ab -c 20 -n 100000 -r http://localhost:5000/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 10000 requests
Completed 20000 requests
Completed 30000 requests
Completed 40000 requests
Completed 50000 requests
Completed 60000 requests
Completed 70000 requests
Completed 80000 requests
Completed 90000 requests
Completed 100000 requests
Finished 100000 requests
Server Software:
Server Hostname: localhost
Server Port: 5000
Document Path: /
Document Length: 11 bytes
Concurrency Level: 20
Time taken for tests: 31.966 seconds
Complete requests: 100000
Failed requests: 351
(Connect: 0, Receive: 2, Length: 179, Exceptions: 170)
Write errors: 0
Total transferred: 11180848 bytes
HTML transferred: 1098119 bytes
Requests per second: 3128.35 [#/sec] (mean)
Time per request: 6.393 [ms] (mean)
Time per request: 0.320 [ms] (mean, across all concurrent requests)
Transfer rate: 341.58 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 3 1.3 3 33
Processing: 0 3 1.6 3 35
Waiting: 0 3 1.5 3 34
Total: 2 6 2.2 6 39
Percentage of the requests served within a certain time (ms)
50% 6
66% 7
75% 7
80% 7
90% 8
95% 9
98% 11
99% 14
100% 39 (longest request)
Raw
send-kill-hup-Starman-SIGQUIT
$ start_server --port 5000 --pid app.pid --signal-on-hup=QUIT --signal-on-term=QUIT -- starman --workers 20 app.psgi
$ ab -c 20 -n 100000 -r http://localhost:5000/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 10000 requests
Completed 20000 requests
Completed 30000 requests
Completed 40000 requests
Completed 50000 requests
Completed 60000 requests
Completed 70000 requests
Completed 80000 requests
Completed 90000 requests
Completed 100000 requests
Finished 100000 requests
Server Software:
Server Hostname: localhost
Server Port: 5000
Document Path: /
Document Length: 11 bytes
Concurrency Level: 20
Time taken for tests: 22.525 seconds
Complete requests: 100000
Failed requests: 240
(Connect: 0, Receive: 0, Length: 120, Exceptions: 120)
Write errors: 0
Total transferred: 11187008 bytes
HTML transferred: 1098724 bytes
Requests per second: 4439.48 [#/sec] (mean)
Time per request: 4.505 [ms] (mean)
Time per request: 0.225 [ms] (mean, across all concurrent requests)
Transfer rate: 485.01 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 2 1.8 2 44
Processing: 0 2 1.9 2 44
Waiting: 0 2 1.7 2 43
Total: 1 4 2.8 4 48
Percentage of the requests served within a certain time (ms)
50% 4
66% 4
75% 5
80% 5
90% 6
95% 6
98% 8
99% 12
100% 48 (longest request)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment