Half of the text was copied from NameResolver.getServiceAuthority().
However, that method can't perform I/O (which would block) so more text
was appropriate here to mention the implications of having a remote
service provide the authority.
I noticed the text was lacking while discussing #9266.
This seems like a poorly thought-out feature for maven_install in a
world where the dependency list is being combined from multiple
repositories, as it means all Bazel-aware dependencies must subscribe to
the idea this is good and useful. It also loses effectiveness as
transitive dependencies are likely to be included directly in
maven_install; users of gRPC can use netty, guava, and other
dependencies without explicitly defining them even with
strict_visibility=True. Given maven_install still complains
about "Found duplicate artifact versions", it isn't too surprising it
doesn't make sense in a multi-repo world.
This isn't a bad feature for gRPC as we are pretty low-level and it can
be helpful when making bazel dependency versions match gradle dependency
versions. But it doesn't necessarily seem like something we need to be
too worried about making sure we use. Similarly, don't use it in the
example as it seems low-value for our users.
Fixes#9288
This Addresses the issue with skips not working due to the
missing/inconsistent `--testing_version` flag, ref b/235688697.
1. Uses the new `TESTING_VERSION` variable populated in the shared
grpc_xds_k8s_install_test_driver.sh - new approach for detecting
versions applicable to all languages.
2. Use `TESTING_VERSION` in all build files in `--testing_version` and
when tagging docker images. This will be backported to all active
test branches. Build Scripts in all other languages will be updated
as well.
Corresponding grpc core change: https://github.com/grpc/grpc/pull/30027
If the failure is before the NameResolver has returned the first time,
RPCs would be queued waiting for service config. We don't want to use
the ConfigSelector, as we are trying to circumvent the NameResolver and
LoadBalancer.
Fixes#9257
This moves our depedencies into a plain file that can be read and
updated by tooling. While the current tooling is not particularly better
than just using gradle-versions-plugin, it should put us on better
footing. gradle-versions-plugin is actually pretty nice, but will be
incompatible with Gradle 8, so we need to wait a bit to see what the
future holds.
Left libraries as an alias for libs to reduce the commit size and make
it easier to revert if we don't end up liking this approach.
We're using Gradle 7.3.3 where it was an incubating fetaure. But in
Gradle 7.4 is became stable.
Internal build failed with
```
grpc/api/src/test/java/io/grpc/GlobalInterceptorsTest.java:28: error:
[JUnit4RunWithMissing] Test class may not be run because it is missing a @RunWith annotation
public class GlobalInterceptorsTest {
^
Did you mean '@RunWith(JUnit4.class)' or 'public abstract class GlobalInterceptorsTest {'?
```
This allows using GrpcCleanupRule with JUnit 5 when combined with
ExternalResourceSupport. We don't really lose anything important when
running with JUnit 4 and this eases migration to JUnit 5.
ExternalResource is now responsible for combining exceptions. after()
cannot throw checked exceptions, so we must now wrap the
InterruptedException. When used with JUnit 5 we are unable to detect the
test failed; we accept that for now but it may be fair to create a new
class for JUnit 5 to be used with `@RegisterExtension` that implements
BeforeEachCallback and AfterTestExecutionCallback to restore the JUnit 4
behavior.
See #5331
Most are not allowed to be zero. Grace period can be zero to immediately
close the connection when the age is hit. A zero permitKeepAliveTime()
simply doesn't enforce any limits on the client.
The choices here precisely matches the pre-existing Netty behavior, but
also seems to make sense in general.
Users appear to be doing `attributes.toString()` to find keys they are
interested in and then unable to find the name of the Key in our API.
They workaround the problem by scanning through `attributes.keys()`
looking for the key of interest. This is an abuse of the keys() API and
unnecessary user friction. They'd happily use the API if they just knew
where to find it.
I added internal to some strings to make it clear that you shouldn't go
looking to use it. There were many strings I didn't change. I focused on
keys most likely to be seen by users, which meant keys in grpc-api and
keys that are available via transport attributes.
See https://github.com/grpc/grpc-java/issues/1764#issuecomment-1139250061
JVM startup costs that happen when dependencies are loaded for the
first time can consume a lot of time (we've occasionally observed
around ~5 seconds of CPU time); this causes frequent test flakes
with xds (google-c2p) when using the current 5 second deadline.
Increasing to 15 seconds should give enough time.
Previously examples-xds depended on the normal hello-world, as it used
the same classes. But since b6601ba273 it has had its own classes and
not had a dependency on `:examples.
* xds: Custom LB configs to support UDPA TypeStruct
The legacy com.github.udpa.udpa.type.v1.TypedStruct proto should be supported in addition to the current com.github.udpa.udpa.type.v1.TypedStruct one.
Co-authored-by: Sergii Tkachenko <hi@sergii.org>
amirhadadi
Wenlong Zhang
sumitd2
Eric Anderson
Sergii Tkachenko
yifeizhuang
Sergii Tkachenko
Kurt Alfred Kluever
larry-safran
sanjaypujare
Terry Wilson
DNVindhya
Feng Li
Jeremy Volkman
Vladimir Chizhov
apolcyn
Lidi Zheng