This removes an abuse of scheduled executor in ManagedChannelImpl. The executor
was used to avoid deadlocking. Now we run the work on the same thread, but
delay it until locks have been released.
There is no need to fix ManagedChannelImpl2. Due to its different
threading model it didn't have need to abuse the scheduledExecutor.
Fixes#2444
Fixes#2207. This is actually a workaround. Ideally users shouldn't need
to -keep classes, but it's a bit risky to fix the real issue before 1.1.
The further fix will be done as part of #2633.
The interop app's build.gradle change is necessary to compile with newer
Gradle versions. The com.google.errorprone.annotations was necessary in
order to prevent annotation warnings from failing the build.
These are only used in internal tests. In production,
StatsContextFactory is loaded by the "Instrumentation" library and must
be one per process, thus we won't allow setting it on a per-channel or
per-server basis.
The interceptors are quite specific, and probably not helpful for
current testing strategies. They really are only useful for interop
testing. Moving them to interop-testing avoids them appearing to be in a
public API (even if that API is experimental).
No functional changes were made; just code movement.
* Remove else, make it easier to read
* Ready subchanns list will be at max the same as the input list
* Make status private, add pkg-private getter for tests
* Remove empty field, cache size, simplify logic
Make list unmodifiable. Document index with GuardedBy (see #nextSubchannel())
add `getAttributes()` to `ClientStream` and `ClientCall` to be able to share clientTransport
information such as socket TOS with higher lever API's, once the RPC picks up an active transport that is ready to use.
Previously it does it at shutdown, which was wrong because executor may
still be used before the server is terminated.
Resolves#2034
Uses ObjectPool to make this change testable. Cleans up test and makes
it mostly single-threaded, except for two deadlock tests that have to be
multi-threaded.
If a LoadBalancer2 is passed in, the builder will create ManagedChannelImpl2 instead of ManagedChannelImpl. This allows us to test the LBv2 classes on a large scale.
Protobuf runtime 3.0.2 is included transitively via grpc-protobuf
v1.0.3. Protoc doesn't have to match the version identically, but must
not be newer than the runtime. We must wait until v1.1 release of grpc
to bump protobuf version in the readme.
Fixes#2578
1. Adapt to LoadBalancer2 interface. Channel holds on to a single
DelayedClientTransport2.
2. Lock-free: every channel state mutation, including Subchannel
mutations, calling into LoadBalancer, idleness and shutdown, is made
from channelExecutor.
3. Idleness grace period is no longer needed.
This patch introduces an additional ALPN protocol, grpc-exp, intended to
take preference to h2 and indicate to the server that the connection
contains only gRPC traffic. This allows servers and intermediate boxes
to distinguish gRPC from other HTTP/2 traffic.
The choice of grpc-exp as a protocol identifier indicates that this
scheme is currently experimental and should not be relied upon. The
protocol is not in the IANA TLS registry.
This is the grpc-java equivalent of
8cdf17a620.
Due to the opacity of ALPN and TLS negotiation at application level, the
tests are only there to validate that the lists we're feeding into the
negotiation process have the desired ordering properties:
* If grpc-exp is present, h2 is as well.
* grpc-exp is preferenced over h2.
In upstream, Census is renamed to "Instrumentation". `com.google.census` is renamed to `com.google.instrumentation.stats`.
In gRPC, "census" in every name is replaced by "stats".
dapengzhang0
Eric Gribkoff
Carl Mastrangelo
Lukasz Strzalkowski
Łukasz Strzałkowski
Eric Anderson
Kun Zhang
Jorg Heymans
Kenji Kaneda
htuch