Previously, mruby context is wrongly shared by multiple patterns if
the underlying SharedDownstreamAddr is shared by multiple
DownstreamAddrGroups. This commit fixes it.
Pool HTTP/1.1 backend connection per address and reuse it only when
the next round robin index refers to this address. Previously if
there is a pooled connection, there is no round robin selection.
If "upgrade-scheme" parameter is present in backend option along with
"tls" paramter, HTTP/2 :scheme pseudo header field is changed to
"https" from "http" when forwarding a request to this particular
backend. This is a workaround for a server which requests "https"
scheme on HTTP/2 connection encrypted by TLS.
This commit removes frontend-tls parameter, and adds
redirect-if-not-tls parameter parameter to --backend option. nghttpx
now responds to the request with 308 status code to redirect the
request to https URI if frontend connection is not TLS encrypted, and
redirect-if-no-tls parameter is used in --backend option. The port
number in Location header field is 443 by default (thus omitted), but
it can be configurable using --redirect-https-port option.
We have added "dns" parameter to backend option. If specified, name
lookup is done dynamically. If not, name lookup is done at start up,
or configuration reloading. nghttpx caches DNS result including error
case in 30 seconds in this commit. Later commit makes this
configurable.
DNS resolution is done asynchronously using c-ares library.
std::make_shared in Xcode 7, 7.1, and 7.2 does not perform
value-initialization, and causes undefined behaviour if struct does
not have user defined default constructor. This workaround explicitly
defines user defined default constructor, and initializes values.
To reduce TTFB with large number of incoming connections, we now
intentionally accept one connection at a time, so that it does not
delay the TTFB of the existing connection. This is significant
especially for TLS connections.
The --backend-tls-sni-field is deprecated in favor of sni keyword.
--backend-tls-sni-field still works, and it overrides all sni keyword
in --backend option. But it will be removed in the future release.
Previously, we use one Http2Session object per DownstreamAddrGroup.
This is not flexible, and we have to provision how many HTTP/2
connection is required in advance. The new strategy is we add
Http2Session object on demand. We measure the number of attached
downstream connection object and server advertised concurrency limit.
As long as former is smaller than the latter, we attach new downstream
connection to it. Once the limit is reached, we create new
Http2Session object. If the number lowers the limit, we start to
share Http2Session object again.
Tatsuhiro Tsujikawa