* Revert "Revert "XdsClient: add unit test and fix watcher notification bugs (#30823)" (#30942)"
This reverts commit 6d2c4a8314.
* use GRPC_CUSTOM_JSONUTIL macro for JsonPrintOptions
This adds a unit test for XdsClient and fixes several watcher-notification bugs found in the process. Specifically:
- When an ADS stream fails or an xDS channel reports a connectivity failure, report an error only to the watchers for resources being subscribed to on that particular channel, not to watchers on other channels.
- Cache the error status for the channel, so that if a new watcher is started after the channel reports the error, we can immediately report that error to the new watcher.
- If a resource is NACKed and has not been previously cached, or does not exist, report that fact to any new watcher that may be started later.
- If a resource in an ADS response is unparseable but is wrapped in a `Resource` wrapper, we do know its name, so record the validation failure in the cache and report it to the watchers.
Co-authored-by: markdroth <markdroth@users.noreply.github.com>
* A modest split of `:gpr` for mpscq support
A full split of the `gpr` target into 25 separate targets is building,
but there are some hurdles to get over with respect to ODR violations
and public API support for the gpr library.
This PR splits off a small chunk of that work.
* Automated change: Fix sanity tests
Co-authored-by: drfloob <drfloob@users.noreply.github.com>
* [chttp2] Tune initial window size/max frame size setting
- Only choose frame sizes and initial window sizes that are powers of two - as a more predictable function to avoid flapping.
- If window size drops below 1kb then snap to 0 to avoid frantically changing a value that needn't be
- Allow the initial window size to drop to zero (we have fuzzing in place that says this is safe)
* copy/paste fix
* fix build
* fix
* Automated change: Fix sanity tests
* fix announce bug
* cleanup
* put this change under an experiment
* more tests
* fix build
* Automated change: Fix sanity tests
Co-authored-by: ctiller <ctiller@users.noreply.github.com>
* client_channel: rewrite illegal status codes from control plane
* rewrite illegal status codes for call creds
* move fail_lb policy out of retry_lb_fail test so it can be reused
* test resolver and LB policy status rewrites
* add test for ConfigSelector status rewriting
* attempt to add client_auth filter unit test
* fix client_auth_filter test
* cleanup test
* fix build
* fix some memory leaks
* Automated change: Fix sanity tests
* Update client_auth_filter_test.cc
* fix build
* code review comments
* clang-tidy
Co-authored-by: markdroth <markdroth@users.noreply.github.com>
Co-authored-by: Craig Tiller <ctiller@google.com>
* [cleanup] Remove profiling timers
- nobody has used this system in years
- if we needed it, we'd probably rewrite it at this point to be something more modern
- let's remove it until that need arises
* fix
* fixes
* Disable end2end_binder_transport_test on some platforms
The following test case is flaky on windows
End2EndBinderTransportTestWithDifferentDelayTimes/End2EndBinderTransportTest.UnaryCallServerTimeout/1,
where GetParam() = 10ns
Binder transport won't be run on platform other than Android so it
should be OK to disable the test on some platform.
* Regenerate projects.
* Reland: "Make GetDefaultEventEngine return a shared_ptr (#30280)"
This reverts commit 45959e7cc1.
* Attempted fix with NoDestruct
* Not a process-wide singleton for the type. Just a NonDestruct
* fix
This works around valgrind memory leaks by giving EventEngines a fixed
lifetime. We eventually want ref-counted EventEngines internally, so this is
a step in the right direction as well.
A (currently) pthread_atfork-based fork support mechanism, allowing EventEngines - or any other object that wants to implement the Forkable interface - respond to forks.
* Rename the default EventEngine headers
Small cleanup. This code hasn't been related to factories for a month or
two.
* ensure only one target contains default_event_engine.h
* src + hdr in same target
* include guards
* [resource_quota] Custom controller
Instead of a PID controller use a custom binary-search inspired controller that will eventually converge to a good control value.
Experiments have shown this able to successfully hold a 95% memory pressure, optimizing for performance most of the time, and gracefully restricting memory usage under severe pressure.
There will be further changes after this one to use the new control value in various systems.
* Automated change: Fix sanity tests
* Update memory_quota.cc
Co-authored-by: ctiller <ctiller@users.noreply.github.com>
* [promises] Eliminate switch.h
Instead build a jump table that we can index into and call the appropriate function. Means that the dispatch mechanism can be done in C++ without a code generator, and so eliminates a bunch of fiddly code.
* fix
* ensure static initialization
* Automated change: Fix sanity tests
Co-authored-by: ctiller <ctiller@users.noreply.github.com>
Craig Tiller
Cheng-Yu Chung
Vignesh Babu
Mark D. Roth
Richard Belleville
Richard Belleville
AJ Heller
Ming-Chuan