URI schema are case-insensitive. Previously the code would do
case-sensitive matching. We expect NameResolverProviders to return the
typical canonical scheme formatting, which is lower-case. If a
NameResolverProvider returns an unexpected string (upper case, unicode,
etc), then it simply won't ever match. Channel users, however, can use
either casing in target strings.
The code implicitly already handled relative URIs by returning null, as
Map.get(null) returned null.
When XdsClient learns that a control plane no longer tracks a resource,
it should only notify watchers associated with that control plane.
This matters in control plane federation cases when more than one
control plane is in use.
This change is to address b/269159638 about GCP metadata server returning an unexpected REFUSED status instead of NXDOMAIN. This causes a 15s delay with older versions of systemd-resolved.
This issue was observed with systemd 245.4. Version 251 worked without issue.
* Add new metric views; Register latency and bytes per metrics views for observability
* update view description to inlcude client/server
* Use pre-defined aggregation from OpenCensus for bytes histogram
* updated view names
Introduce an AsyncService interface in the generated code and move the methods from <service>ImplBase to default implementation of the interface.
* update pom files to allow java 1.8
* Add a bindService(<service>Async) method
* Change TestServiceImpl to use the interface and include a bind method instead of extending TestServiceImplBase.
* Correct value being passed to throttler which had been backwards.
* Fix flaky test.
* Add a test using AdaptiveThrottler with a CachingRlsLBClient.
* Address test flakiness.
* Made dependency on build.gradle in protobuf tasks relative so as to make downstream tasks cacheable
* Made transformer cacheable
* Add property name to path to root build.gradle
Co-authored-by: Nelson Osacky <nelson@osacky.com>
This change has these main aspects to it:
1. Removal of any name resolution responsibility from ManagedChannelImpl
2. Creation of a new RetryScheduler to own generic retry logic
- Can also be used outside the name resolution context
3. Creation of a new RetryingNameScheduler that can be used to wrap any
polling name resolver to add retry capability
4. A new facility in NameResolver to allow implementations to notify
listeners on the success of name resolution attempts
- RetryingNameScheduler relies on this
* xds: allow sum of cluster weights above MAX_INT up to max of unsigned int.
* Define nextLong(long bound) method in FakeRandom for WeightedRandomPickerTest.
Maven seems to have improved their download management and instead of
having their webpage choose a host they now have a CDN domain.
apache.cs.utah.edu is slow and is failing to finish the downloading.
Instead of always overriding the bootstrap with a custom c2p config, now
we allow user defined ones to also be used. This only applies when
running in GCP with federation.
Full end to end implementation of gRPC server as a Servlet including tests and examples
Co-authored-by: Penn (Dapeng) Zhang <zdapeng@google.com>
Co-authored-by: Chengyuan Zhang <chengyuanzhang@google.com>
ExpectedException is deprecated, so I fixed the new warnings. However,
we are still using ExpectedException many places and had previously
supressed the warning. See
https://github.com/grpc/grpc-java/issues/7467 . I did not fix those
existing instances that had suppressed the warning, since it is
unrelated to the upgrade and we have been free to fix them at any time
since we dropped Java 7.
Our tests assume localhost is in /etc/hosts or uses some other form of
local-only resolution. But that wouldn't apply to "host". What was
happening is this was causing a DNS resolution, which would fail, and
the InetSocketAddress would be "unresolved". Thus, the equivalent and
faster code would be `InetSocketAddress.createUnresolved("host", 1234)`.
But there doesn't seem to be any reason to avoid localhost in this test,
so swap to the more typical solution instead.
This should avoid flakes like:
```
io.grpc.okhttp.OkHttpClientTransportTest > invalidAuthorityPropagates FAILED
org.junit.runners.model.TestTimedOutException: test timed out after 10 seconds
at java.base@11.0.17/java.net.Inet6AddressImpl.lookupAllHostAddr(Native Method)
at java.base@11.0.17/java.net.InetAddress$PlatformNameService.lookupAllHostAddr(InetAddress.java:929)
at java.base@11.0.17/java.net.InetAddress.getAddressesFromNameService(InetAddress.java:1529)
at java.base@11.0.17/java.net.InetAddress$NameServiceAddresses.get(InetAddress.java:848)
at java.base@11.0.17/java.net.InetAddress.getAllByName0(InetAddress.java:1519)
at java.base@11.0.17/java.net.InetAddress.getAllByName(InetAddress.java:1378)
at java.base@11.0.17/java.net.InetAddress.getAllByName(InetAddress.java:1306)
at java.base@11.0.17/java.net.InetAddress.getByName(InetAddress.java:1256)
at java.base@11.0.17/java.net.InetSocketAddress.<init>(InetSocketAddress.java:220)
at app//io.grpc.okhttp.OkHttpClientTransportTest.invalidAuthorityPropagates(OkHttpClientTransportTest.java:1687)
```
* Revert "Revert "netty:Auto adjust BDP ping frequency (#9650)" (#9821)"
This reverts commit a2bbe84198.
* Eliminate half RTT delay in sending BDP Pings when starting up.
* Eliminate delay for bdp ping when current read would push us over the threshold.
Apparently there's a difference between bash 3 and bash 4.
OSX comes with bash 3 out-of-box, so for whoever wrote this logic
it "worked on my machine".
The `((` construct returns a 0 exit code if the value is non-zero.
Since the value starts at 0 and we do a post-increment,
it will always fail the first time.
Changing it to a pre-increment should fix it.
Apparently there's a difference between bash 3 and bash 4.
OSX comes with bash 3 out-of-box, so for whoever wrote this logic
it "worked on my machine".
This client can be used as a part of XDS federation integration tests. It can concurrently perform RPCs with different stubs using different underlying XDS servers.
For example, one might perform proxyless service mesh and DirectPath RPCs in the same process with flags like:
```
--server_uris=xds:///${PSM_TARGET},google-c2p:///${DIRECTPATH_TARGET} \
--credentials_types=INSECURE_CREDENTIALS,compute_engine_channel_creds \
--test_case=rpc_soak \
--soak_iterations=10 \
--soak_max_failures=0 \
--soak_per_iteration_max_acceptable_latency_ms=2500 \
--soak_overall_timeout_seconds=300
```
Use a volatile to publish the value even though it is final. TSAN
ignores the final aspect of a field, which is fair since another thread
may not see the parent's pointer become updated and use a stale value.
The lack of synchronization was clearly against lastServiceConfig's
documentation.
Fixes#9267
This helps to prevent retryable stream from using calloptions.executor when it shouldn't, e.g. call is already notified closed. It is done by delaying notifying upper layer (through masterListener).
Fix a bug. When any of the xds subscribers for a resource has the last watcher cancelled, the bug will accidentally remove that resource type from the map, which make xds stream not accepting response update for that resource type entirely(pass through, no ACK/NACK will send).
As normal, Android versions weren't touched as it tends to be special to
upgrade.
The errorprone plugin handles errorproneJavac for us now, since it
hasn't changed in five years. VERSION_CATALOGS is already enabled by
default and graduated out of preview.
Fixes#9802
Eric Anderson
Terry Wilson
yifeizhuang
Larry Safran
DNVindhya
Sergey Matyukevich
Ivan Bahdanau
Asaf Flescher
Kun Zhang
Benjamin Einaudi
chenwei321
Colin Alworth
Alexander Polcyn
pandaapo
Sergii Tkachenko