view http_header_buffers.t @ 1752:ba6e24e38f03

Tests: improved stop_daemons() to send signal again. As was observed, it's possible that a signal to complete a uwsgi daemon can be ignored while it is starting up, which results in tests hang due to eternal waiting on child processes termination. Notably, it is seen when running tests with a high number of prove jobs on a low-profile VM against nginx with broken modules and/or configuration. To reproduce: $ TEST_NGINX_GLOBALS=ERROR prove -j16 uwsgi*.t Inspecting uwsgi under ktrace on FreeBSD confirms that a SIGTERM signal is ignored at the very beginning of uwsgi startup. It is then replaced with a default action after listen(), thus waiting until uwsgi is ready to accept new TCP connections doesn't completely solve the hang window. The fix is to retry sending a signal some time after waitpid(WNOHANG) continuously demonstrated no progress with reaping a signaled process. It is modelled after f13ead27f89c that improved stop() for nginx.
author Sergey Kandaurov <pluknet@nginx.com>
date Wed, 29 Dec 2021 22:29:23 +0300
parents 66c7dee0431c
children
line wrap: on
line source

#!/usr/bin/perl

# (C) Maxim Dounin
# (C) Nginx, Inc.

# Tests for large_client_header_buffers directive.

###############################################################################

use warnings;
use strict;

use Test::More;

use Socket qw/ CRLF /;

BEGIN { use FindBin; chdir($FindBin::Bin); }

use lib 'lib';
use Test::Nginx;

###############################################################################

select STDERR; $| = 1;
select STDOUT; $| = 1;

my $t = Test::Nginx->new()->has(qw/http rewrite/)->plan(10)
	->write_file_expand('nginx.conf', <<'EOF');

%%TEST_GLOBALS%%

daemon off;

events {
}

http {
    %%TEST_GLOBALS_HTTP%%

    connection_pool_size 128;
    client_header_buffer_size 128;

    server {
        listen       127.0.0.1:8080;
        server_name  five;

        large_client_header_buffers 5 256;

        return 204;
    }

    server {
        listen       127.0.0.1:8080;
        server_name  ten;

        large_client_header_buffers 10 256;

        return 204;
    }

    server {
        listen       127.0.0.1:8080;
        server_name  one;

        large_client_header_buffers 1 256;

        return 204;
    }

    server {
        listen       127.0.0.1:8080;
        server_name  foo;

        large_client_header_buffers 5 256;

        add_header X-URI $uri;
        add_header X-Foo $http_x_foo;
        return 204;
    }
}

EOF

$t->run();

###############################################################################

TODO: {
todo_skip 'overflow', 2 unless $ENV{TEST_NGINX_UNSAFE};

# if hc->busy is allocated before the virtual server is selected,
# and then additional buffers are allocated in a virtual server with larger
# number of buffers configured, hc->busy will be overflowed

like(http(
	"GET / HTTP/1.0" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"Host: ten" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	CRLF
), qr/204|400/, 'additional buffers in virtual server');

# for pipelined requests large header buffers are saved to hc->free;
# it sized for number of buffers in the current virtual server, but
# saves previously allocated buffers, and there may be more buffers if
# allocatad before the virtual server was selected

like(http(
	"GET / HTTP/1.1" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"Host: one" . CRLF .
	CRLF .
	"GET / HTTP/1.1" . CRLF .
	"Host: one" . CRLF .
	"Connection: close" . CRLF .
	CRLF
), qr/204/, 'pipelined with too many buffers');

}

# check if long header and long request lines are correctly returned
# when nginx allocates a long header buffer

like(http(
	"GET / HTTP/1.0" . CRLF .
	"Host: foo" . CRLF .
	"X-Foo: foo" . ("1234567890" x 20) . "bar" . CRLF .
	CRLF
), qr/X-Foo: foo(1234567890){20}bar/, 'long header');

like(http(
	"GET /foo" . ("1234567890" x 20) . "bar HTTP/1.0" . CRLF .
	"Host: foo" . CRLF .
	CRLF
), qr!X-URI: /foo(1234567890){20}bar!, 'long request line');

# the same as the above, but with pipelining, so there is a buffer
# allocated in the previous request

like(http(
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	CRLF .
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"Connection: close" . CRLF .
	"X-Foo: foo" . ("1234567890" x 20) . "bar" . CRLF .
	CRLF
), qr/X-Foo: foo(1234567890){20}bar/, 'long header after pipelining');

like(http(
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	CRLF .
	"GET /foo" . ("1234567890" x 20) . "bar HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"Connection: close" . CRLF .
	CRLF
), qr!X-URI: /foo(1234567890){20}bar!, 'long request line after pipelining');

# the same as the above, but with keepalive; this ensures that previously
# allocated buffers are properly cleaned up when we set keepalive handler

like(http(
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	CRLF,
sleep => 0.1, body =>
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"Connection: close" . CRLF .
	"X-Foo: foo" . ("1234567890" x 20) . "bar" . CRLF .
	CRLF
), qr/X-Foo: foo(1234567890){20}bar/, 'long header after keepalive');

like(http(
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	CRLF,
sleep => 0.1, body =>
	"GET /foo" . ("1234567890" x 20) . "bar HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"Connection: close" . CRLF .
	CRLF
), qr!X-URI: /foo(1234567890){20}bar!, 'long request line after keepalive');

# the same as the above, but with pipelining and then keepalive;
# this ensures that previously allocated buffers are properly cleaned
# up when we set keepalive handler, including hc->free

like(http(
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	CRLF .
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	CRLF,
sleep => 0.1, body =>
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"Connection: close" . CRLF .
	"X-Foo: foo" . ("1234567890" x 20) . "bar" . CRLF .
	CRLF
), qr/X-Foo: foo(1234567890){20}bar/, 'long header after both');

like(http(
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	CRLF .
	"GET / HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"X-Foo: " . ("1234567890" x 20) . CRLF .
	CRLF,
sleep => 0.1, body =>
	"GET /foo" . ("1234567890" x 20) . "bar HTTP/1.1" . CRLF .
	"Host: foo" . CRLF .
	"Connection: close" . CRLF .
	CRLF
), qr!X-URI: /foo(1234567890){20}bar!, 'long request line after both');

###############################################################################