Iterating through all connections takes a lot of CPU time, especially
with large number of worker connections configured. As a result
nginx processes used to consume CPU time during graceful shutdown.
To mitigate this we now only do a full scan for idle connections when
shutdown signal is received.
Transitions of connections to idle ones are now expected to be
avoided if the ngx_exiting flag is set. The upstream keepalive module
was modified to follow this.
When configured, an individual listen socket on a given address is
created for each worker process. This allows to reduce in-kernel lock
contention on configurations with high accept rates, resulting in better
performance. As of now it works on Linux and DragonFly BSD.
Note that on Linux incoming connection requests are currently tied up
to a specific listen socket, and if some sockets are closed, connection
requests will be reset, see https://lwn.net/Articles/542629/. With
nginx, this may happen if the number of worker processes is reduced.
There is no such problem on DragonFly BSD.
Based on previous work by Sepherosa Ziehau and Yingqi Lu.
This reduces layering violation and simplifies the logic of AIO preread, since
it's now triggered by the send chain function itself without falling back to
the copy filter. The context of AIO operation is now stored per file buffer,
which makes it possible to properly handle cases when multiple buffers come
from different locations, each with its own configuration.
Client address specified in the PROXY protocol header is now
saved in the $proxy_protocol_addr variable and can be used in
the realip module.
This is currently not implemented for mail.
Fallback to synchronous sendfile() now only done on 3rd EBUSY without
any progress in a row. Not falling back is believed to be better
in case of occasional EBUSY, though protection is still needed to
make sure there will be no infinite loop.
The c->single_connection was intended to be used as lock mechanism
to serialize modifications of request object from several threads
working with client and upstream connections. The flag is redundant
since threads in nginx have never been used that way.
There is a general consensus that this change results in better
consistency between different operating systems and differently
tuned operating systems.
Note: this changes the width and meaning of the ipv6only field
of the ngx_listening_t structure. 3rd party modules that create
their own listening sockets might need fixing.
*) Feature: the "bind" option of the "listen" directive in IMAP/POP3
proxy.
*) Bugfix: if the same capture in the "rewrite" directive was used more
then once.
*) Bugfix: the $sent_http_content_type, $sent_http_content_length,
$sent_http_last_modified, $sent_http_connection,
$sent_http_keep_alive, and $sent_http_transfer_encoding variables
were not written to access log.
*) Bugfix: the $sent_http_cache_control returned value of the single
"Cache-Control" response header line.
*) Feature: the IMAP/POP3 proxy supports STARTTLS and STLS.
*) Bugfix: the IMAP/POP3 proxy did not work with the select, poll, and
/dev/poll methods.
*) Bugfix: in SSI handling.
*) Bugfix: now Solaris sendfilev() is not used to transfer the client
request body to FastCGI-server via the unix domain socket.
*) Bugfix: the "auth_basic" directive did not disable the
authorization; the bug had appeared in 0.3.11.
*) Change: the "valid_referers" directive and the "$invalid_referer"
variable were moved to the new ngx_http_referer_module from the
ngx_http_rewrite_module.
*) Change: the "$apache_bytes_sent" variable name was changed to
"$body_bytes_sent".
*) Feature: the "$sent_http_..." variables.
*) Feature: the "if" directive supports the "=" and "!=" operations.
*) Feature: the "proxy_pass" directive supports the HTTPS protocol.
*) Feature: the "proxy_set_body" directive.
*) Feature: the "post_action" directive.
*) Feature: the ngx_http_empty_gif_module.
*) Feature: the "worker_cpu_affinity" directive for Linux.
*) Bugfix: the "rewrite" directive did not unescape URI part in
redirect, now it is unescaped except the %00-%25 and %7F-%FF
characters.
*) Bugfix: nginx could not be built by the icc 9.0 compiler.
*) Bugfix: if the SSI was enabled for zero size static file, then the
chunked response was encoded incorrectly.
*) Change: the "bl" and "af" parameters of the "listen" directive was
renamed to the "backlog" and "accept_filter".
*) Feature: the "rcvbuf" and "sndbuf" parameters of the "listen"
directive.
*) Change: the "$msec" log parameter does not require now the
additional the gettimeofday() system call.
*) Feature: the -t switch now tests the "listen" directives.
*) Bugfix: if the invalid address was specified in the "listen"
directive, then after the -HUP signal nginx left an open socket in
the CLOSED state.
*) Bugfix: the mime type may be incorrectly set to default value for
index file with variable in the name; the bug had appeared in 0.3.0.
*) Feature: the "timer_resolution" directive.
*) Feature: the millisecond "$upstream_response_time" log parameter.
*) Bugfix: a temporary file with client request body now is removed
just after the response header was transferred to a client.
*) Bugfix: OpenSSL 0.9.6 compatibility.
*) Bugfix: the SSL certificate and key file paths could not be relative.
*) Bugfix: the "ssl_prefer_server_ciphers" directive did not work in
the ngx_imap_ssl_module.
*) Bugfix: the "ssl_protocols" directive allowed to specify the single
protocol only.
*) Feature: the "config errmsg" command of the ngx_http_ssi_module.
*) Change: the ngx_http_geo_module variables can be overridden by the
"set" directive.
*) Feature: the "ssl_protocols" and "ssl_prefer_server_ciphers"
directives of the ngx_http_ssl_module and ngx_imap_ssl_module.
*) Bugfix: the ngx_http_autoindex_module did not show correctly the
long file names;
*) Bugfix: the ngx_http_autoindex_module now do not show the files
starting by dot.
*) Bugfix: if the SSL handshake failed then another connection may be
closed too.
Thanks to Rob Mueller.
*) Bugfix: the export versions of MSIE 5.x could not connect via HTTPS.
*) The pid-file names used during online upgrade was changed and now is
not required a manual rename operation. The old master process adds
the ".oldbin" suffix to its pid-file and executes a new binary file.
The new master process creates usual pid-file without the ".newbin"
suffix. If the master process exits, then old master process renames
back its pid-file with the ".oldbin" suffix to the pid-file without
suffix.
*) Change: the "worker_connections" directive, new name of the
"connections" directive; now the directive specifies maximum number
of connections, but not maximum socket descriptor number.
*) Feature: SSL supports the session cache inside one worker process.
*) Feature: the "satisfy_any" directive.
*) Change: the ngx_http_access_module and ngx_http_auth_basic_module do
not run for subrequests.
*) Feature: the "worker_rlimit_nofile" and "worker_rlimit_sigpending"
directives.
*) Bugfix: if all backend using in load-balancing failed after one
error, then nginx did not try do connect to them during 60 seconds.
*) Bugfix: in IMAP/POP3 command argument parsing.
Thanks to Rob Mueller.
*) Bugfix: errors while using SSL in IMAP/POP3 proxy.
*) Bugfix: errors while using SSI and gzipping.
*) Bugfix: the "Expires" and "Cache-Control" header lines were omitted
from the 304 responses.
Thanks to Alexandr Kukushkin.
*) Feature: the IMAP/POP3 proxy supports SSL.
*) Feature: the "proxy_timeout" directive of the ngx_imap_proxy_module.
*) Feature: the "userid_mark" directive.
*) Feature: the $remote_user variable value is determined independently
of authorization use.
*) Feature: the listen(2) backlog in the "listen" directive can be
changed using the -HUP signal.
*) Feature: the geo2nginx.pl script was added to contrib.
*) Change: the FastCGI parameters with the empty values now are passed
to a server.
*) Bugfix: the segmentation fault occurred or the worker process may
got caught in an endless loop if the proxied or FastCGI server sent
the "Cache-Control" header line and the "expires" directive was
used; in the proxied mode the the bug had appeared in 0.1.29.
*) Feature: the "limit_rate" directive is supported in in proxy and
FastCGI mode.
*) Feature: the "X-Accel-Limit-Rate" response header line is supported
in proxy and FastCGI mode.
*) Feature: the "break" directive.
*) Feature: the "log_not_found" directive.
*) Bugfix: the response status code was not changed when request was
redirected by the ""X-Accel-Redirect" header line.
*) Bugfix: the variables set by the "set" directive could not be used
in SSI.
*) Bugfix: the segmentation fault may occurred if the SSI page has more
than one remote subrequest.
*) Bugfix: nginx treated the backend response as invalid if the status
line in the header was transferred in two packets; the bug had
appeared in 0.1.29.
*) Feature: the "ssi_types" directive.
*) Feature: the "autoindex_exact_size" directive.
*) Bugfix: the ngx_http_autoindex_module did not support the long file
names in UTF-8.
*) Feature: the IMAP/POP3 proxy.
*) Change: if the request header has duplicate the "Host",
"Connection", "Content-Length", or "Authorization" lines, then nginx
now returns the 400 error.
*) Change: the "post_accept_timeout" directive was canceled.
*) Feature: the "default", "af=", "bl=", "deferred", and "bind"
parameters of the "listen" directive.
*) Feature: the FreeBSD accept filters support.
*) Feature: the Linux TCP_DEFER_ACCEPT support.
*) Bugfix: the ngx_http_autoindex_module did not support the file names
in UTF-8.
*) Bugfix: the new log file can be rotated by the -USR1 signal only if
the reconfiguration by the -HUP signal was made twice.
*) Feature: the "working_directory" directive.
*) Feature: the "port_in_redirect" directive.
*) Bugfix: the segmentation fault was occurred if the backend response
header was in several packets; the bug had appeared in 0.1.29.
*) Bugfix: if more than 10 servers were configured or some server did
not use the "listen" directive, then the segmentation fault was
occurred on the start.
*) Bugfix: the segmentation fault might occur if the response was
bigger than the temporary file.
*) Bugfix: nginx returned the 400 response on requests like
"GET http://www.domain.com/uri HTTP/1.0"; the bug had appeared in
0.1.28.
*) Bugfix: nginx could not be built with the --without-pcre parameter;
the bug had appeared in 0.1.29.
*) Bugfix: 3, 4, 7, and 8 the "proxy_set_header" directives in one
level cause the bus fault on start up.
*) Bugfix: the HTTP protocol was specified in the HTTPS redirects.
*) Bugfix: if the "rewrite" directive used the captures inside the "if"
directive, then the 500 error code was returned.
*) Bugfix: the ngx_http_stub_status_module showed incorrect handled
connections statistics if the proxying or FastCGI server were used.
*) Bugfix: the installation paths were incorrectly quoted on Linux and
Solaris; the bug had appeared in 0.1.21.
*) Bugfix: the ngx_http_stub_status_module showed incorrect statistics
if "rtsig" method was used or if several worker process ran on SMP.
*) Bugfix: nginx could not be built by the icc compiler on Linux or if
the zlib-1.2.x library was building from sources.
*) Bugfix: nginx could not be built on NetBSD 2.0.
