Reimplemented close/idle_timeout/stop/onGoAway/input_shutdown following more closely the specification.
In particular, the semantic of sending a GOAWAY is now to:
* stop creation of new both local and remote streams
* record the last processed stream
* continue processing streams that are pending
This means that a GOAWAY is "graceful" in the sense that it allows for streams to be completed by applications.
The semantic of stop() and idle timeout is harsher: for pending streams a RST_STREAM is sent to the other peer and they are failed locally.
Added support for GOAWAY with 2^31-1 lastStreamId.
Added support for a peer to send and receive multiple GOAWAY frames.
Reviewed the stream creation/destruction mechanism so that when the last stream completes after a GOAWAY, proper actions can be run to tear down the connection.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Fixes#3766 - Introduce HTTP/2 API to batch frames.
Introduced Stream.FrameList to hold HEADERS+DATA+HEADERS frames.
These are often used by the client and by the server when the
request/response content is known and FrameList will allow to
send them in a single TCP write, rather than multiple ones.
Rewritten HttpSenderOverHTTP2.sendHeaders() and
HttpTransportOverHTTP2.sendHeaders() to take advantage of
FrameList.
Now using ConcurrentHashMap as a client context, because
with DEBUG logging enabled it may be access concurrently.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Reworked HTTP/2 release after an exchange is terminated.
Previously, the release was bound to 2 events: onStreamClosed(),
introduced for #2796, and exchangeTerminated().
Unfortunately, if the former happens before the latter and
closes the connection, the latter will see the exchange as
aborted, while in fact it was successful, causing what
reported in #5147, an AsynchronousCloseException.
Now, the release is always performed by the exchangeTerminated()
event. With respect to #2796, the stream is always already
closed by the time the exchangeTerminated() event fires (it
was not before).
Reworked the implementation of RoundRobinConnectionPool using
a lock and aggressively trying to open new connections.
A second fix is related to HttpDestination.release(Connection).
If the connection is closed for e.g. overuse, we need to trigger
the processing of queued requests via send(create: true).
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Replaced relevant usages of synchronized with AutoLock.
* Made AutoLock serializable since classes that use it may be stored in the HttpSession.
* Added convenience methods to AutoLock to execute lambdas with the lock held.
* Introduced AutoLock.WithCondition to use a Lock and a Condition together.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Fixed MaxConcurrentStreamsTest - it was always broken.
The problem was that the call to super.onSettings(...) was done
_after_ sending the request, so the connection pool was still
configured with the default maxMultiplex=1024.
Also fixed AbstractConnectionPool to avoid a second call to
activate() if we are not trying to create a new connection.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Fixes#4855 - Occasional h2spec failures on CI
In case of bad usage of the HTTP/2 API, we don't want to close()
the stream but just fail the callback, because the stream
may be performing actions triggered by a legit API usage.
In case of a call to `AsyncListener.onError()`, applications may decide to call
AsyncContext.complete() and that would be a correct usage of the Servlet API.
This case was not well handled and was wrongly producing a WARN log with an
`IllegalStateException`.
Completely rewritten `HttpTransportOverHTTP2.TransportCallback`.
The rewrite handles correctly asynchronous failures that now are executed
sequentially (and not concurrently) with writes.
If a write is in progress, the failure will just change the state and at the
end of the write a check on the state will determine what actions to take.
A session failure is now handled in HTTP2Session by first failing all the
streams - which notifies the Stream.Listeners - and then failing the session
- which notifies the Session.Listener.
The stream failures are executed concurrently by dispatching each one to a
different thread; this means that the stream failure callbacks are executed
concurrently (likely sending RST_STREAM frames).
The session failure callback is completed only when all the stream failure
callbacks have completed, to ensure that a GOAWAY frame is processed after
all the RST_STREAM frames.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Fixes#4904 - WebsocketClient creates more connections than needed.
Fixed connection pool's `acquire()` methods to correctly take into account the number of queued requests.
Now the connection creation is conditional, triggered by
explicit send() or failures.
The connection creation is not triggered _after_ a send(),
where we aggressively send more queued requests - or
in release(), where we send queued request after a previous
one was completed.
Now the connection close/removal aggressively sends more
requests triggering the connection creation.
Also fixed a collateral bug in `BufferingResponseListener` - wrong calculation of the max content length.
Restored `ConnectionPoolTest` that was disabled in #2540, cleaned it up, and let it run for hours without failures.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Introduced:
* Request Request.headers(Consumer<HttpFields.Mutable>).
This allows applications to modify the headers, and chain calls.
It also delegates the precise semantic of put/add/remove/clear to HttpFields, so there is no API duplication.
* HttpRequest.header(HttpField) to efficiently add fields while normalizing the request (only used in implementation).
* HttpResponse.header(HttpField) to efficiently add fields while parsing the response (only used in implementation).
This pairs with HttpResponse.trailer(HttpField).
* HttpResponse.headers(Consumer<HttpFields.Mutable>) to modify the fields after they have been populated (only used in tests).
Removed:
* Request.[set,add,put,remove], replaced by headers(Consumer<HttpFields.Mutable>).
Deprecated:
* Request.header(String, String)
* Request.header(HttpHeader, String)
Both replaced by headers(Consumer<HttpFields.Mutable>) with clearer semantic for add/put/remove.
All the rest is code cleanup to remove the usage of the deprecated header() methods.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Made HttpURI, HttpFields and MetaData immutable. The first two follow the same builder pattern and MetaData is constructor injection only.
* Immutable version of HttpFields
Preserve API and usage of HttpFields class while providing a read only interface and immutable implementation.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable version of HttpFields
Use an ArrayList in HttpFields. While slightly slower than the array, it will mostly be used as a builder pattern for an Immutable
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable version of HttpFields
Fixed exception type.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable version of HttpFields
asImmutable method
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
Made HttpURIU immutable with a builder pattern.
MetaData immutable and working within http module.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
Fixes from review
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
Passing tests upto and including jetty-server
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
Cleanup of HttpURI.Builder API as suggested in PR.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
Added builder for MetaData.Request
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
more api fixes
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
WIP making HttpFiels itself immutable. Currently working up to jetty-servlet.
Need to consider if content-length really is meta data and how much and when can we trust it.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
WIP
Need to consider if content-length really is meta data and how much and when can we trust it. Also need to consider difference between h2 and h1 authority in metadata.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
jetty-client and jetty-servlet passing tests.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
Better align the style of immutability between `HttpFields` and `HttpURI`.
They both now have static build() and from() methods, plus Builder and Immutable implementations.
Potentially `Builder` could be renamed as `Mutable`
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
http2-server tests passed
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
http2-client tests passed
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
cleann build?
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
fix
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
more test fixes
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
Cleanups, mostly using EMPTY when appropriate.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
Cleanups, use immutable
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
No trailers for connect
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
Fix CONNECT path handling
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
fixed rewrite query handling
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
rename Builders to Muttables
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
misc cleanups
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
Revert to using arrays due to garbage generated by streams and iterators (12% of a simple benchmark!).
Even if this garbage is an artifact of the JIT being disabled by observation, it can hide other allocations, so best to just use simple arrays!
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
More optimizations and better test coverage.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable Metadata
various cleanups
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
More optimizations
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
review changes
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
changes after review:
+ less usage of Mutable
+ more usage of EMPTY
+ restored fragment handling
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
changes after review:
+ less usage of Mutable
+ less usage of asImmutable
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData WIP
changes after review:
+ less usage of Mutable
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
changes after review:
+ better handling of URI in ContextHandler
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
changes after review:
+ downcast in test to access mutable response headers.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Immutable MetaData
changes after review:
+ use put instead of add for one time headers
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* private
Signed-off-by: Greg Wilkins <gregw@webtide.com>
Modified jetty-alpn-openjdk8-* classes to support both
pre 8u252 (via alpn-boot) and post 8u252 (via standard API).
Replaced usages of -Xbootclasspath with -javaagent, and
using Jetty ALPN Agent jar rather than Jetty ALPN boot jar.
Removed all alpn-1.8.0*.mod files since now it is
possible to use a fixed version of the ALPN Agent
to cover all the versions.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Introducing jetty-slf4j-impl
* Make Jetty use org.slf4j
* Removed most of org.eclipse.jetty.util.log classes
* Left org.eclipse.jetty.util.log.Log and
org.eclipse.jetty.util.log.Logger but as
simple bridge classes that are deprecated
* Migrated code using org.eclipse.jetty.util.log.StacklessLogging
to org.eclipse.jetty.logging.StacklessLogging found in
the jetty-slf4j-impl
* Moved logging start modules from jetty-util to jetty-home
* Simplified logging start modules
* Updated code that was using StdErrLog directly
* Updating module-info.java for org.slf4j
* removing org.eclipse.jetty.util.log.class references
* jetty-start supports manually declared default provider
+ and we use it to default "logging" to the "logging-jetty" provider
* Cleaning up jetty-maven-plugin and IT testing for Logging
* Using old slf4j for it testing
* Updating compiler config to show Xlint:exports warnings
* Updating console-capture and logging-noop
* Adding slf4j bridge (capture) jetty modules
* Updates to jetty logging module locations
* Changing reference to slf4j dependent mod
* Process requested enabled modules in topological order
* Limiting inclusions in shaded jetty-start
+ Also adding note to jetty-util classes that are used by
jetty-start
* Default logging level on baseline logging config is INFO (not DEBUG)
* Changing from system to server classes in logging
* Updating other modules to use new logging names
Signed-off-by: Joakim Erdfelt <joakim.erdfelt@gmail.com>
* Moved implementation of UpgradeTo from HTTP2ServerConnection
to HTTP2Connection, since now also the client connection
can be upgraded to.
* Split HTTP2Session.newStream(), since now the client must
be able to create the implicit stream 1 corresponding to
the HTTP/1.1 upgrade request, so that the HTTP/2 response
after the upgrade finds the stream.
* The HTTP/1.1 upgrade mechanism has been generalized.
Before it was based on HttpConnectionUpgrader and a hook
in HttpChannelOverHTTP.exchangeTerminating().
Now it is based on UpgradeProtocolHandler that when sees
a 101 response it will trigger the upgrade mechanism.
* Introduced ConnectionPool.accept(Connection) to transfer
a connection from the HTTP/1.1 connection pool to the
HTTP/2 connection pool after the upgrade.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Introduced Request.Content with a reactive model to provide
request content.
Introduced RequestContentAdapter to wrap ContentProviders
into Request.Content.
Updated implementation to use the reactive model rather than
the old pull model.
Reimplemented all ContentProviders in terms of Request.Content.
Converted most of the tests from ContentProvider to Request.Content.
Updated proxy servlets and documentation.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Simone Bordet
Ludovic Orban
Olivier Lamy
Jesse McConnell
Joakim Erdfelt
Lachlan Roberts
Jan Bartel
Greg Wilkins