shrpx:
* Added an option to set the TLS SNI extension between shrpx and the
origin on the command line
spdycat:
* If the user set an explicit host header ( using --headers ) use that
name for the TLS SNI extension.
* Added the handshake completion time to the verbose output
* The gettimeofday call in get_time was using the incorrect structure
( I believe )
* In update_html_parser it was submitting the request regardless of
the return value of add_request.
Patch from Stephen Ludin
spdy_data_read_callback in SpdyDownstreamConnection calls
SpdyUpstream::resume_read() which submits WINDOW_UPDATE, but after
that they are not call SpdyUpstream::send(). This means that if no
pending outgoing data in upstream, then WINDOW_UPDATE is blocked until
SpdyUpstream::send() from somewhere. This change adds
SpdyUpstream::send() to resume_read() so that WINDOW_UPDATE is not
blocked.
This change fixes upstream RST_STREAM is blocked until
SpdyUpstream::send() is called. Now downstream REFUSED_STREAM is
propagated to upstream client so that client can reset request. The
RST_STREAM error code when downstream went wrong is changed from
CANCEL to INTERNAL_ERROR.
Now we set Downstream::set_response_connection_close(true) for
tunneled connections. Also call
Upstream::on_downstream_body_complete() callback when setting
MSG_COMPLETE in SpdySession when RST_STREAM is caught. Clean up EOF
handling in https_downstream_readcb.
Currently, resume_read() fails if on_read() returns -1 in case that
evbuffer_add failed, which means, most likely, memory allocation
failure. ClientHandler is marked "should be closed", but if
evbuffer_add is failed, write callback will not be invoked and its
marking is not evaluated. It will eventually be deleted when the
client is disconnected or backend failure though.
Specify proxy URI in the form http://[USER:PASS]PROXY:PORT. USER and
PASS are optional and if they exist they must be properly
percent-encoded. This proxy is used when the backend connection is
SPDY. First, make a CONNECT request to the proxy and it connects to
the backend on behalf of shrpx. This forms tunnel. After that, shrpx
performs SSL/TLS handshake with the downstream through the tunnel. The
timeouts when connecting and making CONNECT request can be specified
by --backend-read-timeout and --backend-write-timeout options.
With --spdy-bridge option, it listens SPDY/HTTPS connections from
front end and forwards them to the backend in SPDY. The usage will be
written later. This change fixes the crash when more than 2
outstanding SpdyDownstreamConnection objects are added to SpdySession
and establishing connection to SPDY backend is failed.
This option specifies additional certificate and private key
file. Shrpx will choose certificates based on the hostname indicated
by client using TLS SNI extension. This option can be used multiple
times.
INFO log and its surrounding code are now guarded by
LOG_ENABLED(SEVERITY) macro so that they don't run if log level
threshold is higher. This increases performance because log formatting
is somewhat expensive.
Added macros which log messages from the following components are
prefixed with their component name + object pointer address:
ListenHandler: LISTEN
ThreadEventReceiver: THREAD_RECV
Upstream: UPSTREAM
Downstream: DOWNSTREAM
DownstreamConnection: DCONN
SpdySession: DSPDY
This avoids the need to provide the password for your
private key interactively.
It can be used via --private-key-passwd-file or private-key-passwd-file
in the given config file. The first line in the file
(without \n) will be treated as the passwd. There isn't
any validation and all lines after the first one (if any)
are ignored.
The security model behind this is a bit simplistic so I
am open to better ideas. Basically your password file
should be root:root (700) and you *should* drop root
and run as an unprivileged user.
If the file exists and a line can be read then a callback
will be set for the SSL ctxt and it'll feed the passwd
when the private key is read (if password is needed).
If the file exists with the wrong permisions it'll be
logged and ignored.
The -k, --insecure option is added to skip this verification. The
system wide trusted CA certificates will be loaded at startup. The
--cacert option is added to specify the trusted CA certificate file.
In client mode, now SPDY connection to the backend server is
established per thread. The frontend connections which belong to the
same thread share the SPDY connection.
With --client-mode option, shrpx now accepts unencrypted HTTP
connections and communicates with backend server in SPDY. In short,
this is the "reversed" operation mode against normal mode. This may
be useful for testing purpose because it can sit between HTTP client
and shrpx "normal" mode.
We should only call daemon() after ListenHandler is
instantiated, where SSL_CTX_use_PrivateKey_file is called,
otherwise we have no stdin/stdout to get the password for
keyfile.
The reason why we choose SPDY/2 as default for SPDY prxy was due to
Chrome's window update bug. Now its fix is available in Chrome stable,
we make SPDY/3 as default.
The on_ctrl_not_send_callback, on_ctrl_recv_parse_error_callback and
on_unknown_ctrl_recv_callback were added. The latter 2 callbacks are
purely debugging purpose. In on_ctrl_not_send_callback, If sending
SYN_REPLY failed, issue RST_STREAM to avoid a stream hanging around.
When deciding whether to close the client connection, check
request_connection_close_ of Downstream in addition of
response_connection_close_. Also we only add "Connection: Keep-Alive"
header to the HTTP/1.0 or HTTP/0.9 clients.
When --enable-src is given, the programs in src directory will be
built. If --disable-src is given, those programs will not be built. If
none of them are given, --enable-src is assumed.
To distinguish the to-be-installed programs and non-installable
example source code, the former programs, spdycat, spdydyd and shrpx,
were moved to src directory. spdynative was removed from Makefile
because it does not appeal to any users much.
Tatsuhiro Tsujikawa
moparisthebest
snnn
Raul Gutierrez Segales