Causes crashes during shutdown.
<!--
If you know who should review your pull request, please assign it to that
person, otherwise the pull request would get assigned randomly.
If your pull request is for a specific language, please add the appropriate
lang label.
-->
Closes#37195
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/37195 from yousukseung:work-serializer-dispatch-disable 781bc59384
PiperOrigin-RevId: 650741850
Instead of passing the transport byte counts back up through the filter
stack to be reported to the `CallTracer`, we now have the transport
pass the transport byte counts directly to the `CallTracer` itself.
This will eventually allow us to avoid unnecessarily storing these byte
counts in cases where no `CallTracer` actually cares about the data, which
will reduce per-call memory. (In the short term, it actually increases
memory usage, but we can separately do some work to avoid the memory
usage in the transport by removing the `grpc_transport_stream_stats`
struct from the legacy filter API.)
This is a prereq for supporting `CallTracer` in the new call v3 stack,
which does not include the transport byte counts as part of the
receieve-trailing-metadata hook, unlike the legacy filter stack.
This change is controlled by the `call_tracer_in_transport` experiment,
which is enabled by default.
As part of this experiment, we also fix a couple of related bugs:
- On the client side, the chttp2 transport was incorrectly adding
annotations to the parent `ClientCallTracer` instead of the
`CallAttemptTracer`.
- The OpenCensus `ServerCallTracer` was incorrectly swapping the values
of sent and received bytes.
PiperOrigin-RevId: 650728181
This allows CallTracers to be created with parameters dictated by
channel args.
For the moment, I've used the EventEngine `EndpointConfig` API to expose
the channel args here, so as to avoid directly exposing
`grpc_core::ChannelArgs`. We should determine a better API here before
we de-experimentalize the stats APIs.
Also add an experiment to be used in a subsequent PR.
PiperOrigin-RevId: 647730284
Reattempt enabling the Windows EventEngine client experiment, after fixing a race condition in connection establishment vs timeout.
Closes#36876
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/36876 from drfloob:wineec 125f3a20d8
PiperOrigin-RevId: 642294760
Implements https://github.com/grpc/proposal/pull/429
Currently, the behavior of `GRPC_ARG_HTTP2_MAX_PINGS_WITHOUT_DATA` blocks more pings from being sent if we are sending too many pings without a data/header frame being sent as well. The original intention of this channel arg was to play nice with proxies that have restrictive settings when it comes to pings. This causes awkwardness when configuring keepalive pings for transports with long lived streams with sparse communication. In such a case, gRPC Core would stop sending keepalive pings since no data/header frame is being sent, resulting in a situation where we are unable to detect whether the transport is alive or not.
This change adds an experiment "max_pings_wo_data_throttle" to modify the behavior of `GRPC_ARG_HTTP2_MAX_PINGS_WITHOUT_DATA` to throttle pings to a frequency of 1 minute instead of completely blocking pings when too many pings have been sent without data/header frames.
Closes#36374
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/36374 from yashykt:ThrottlePings b5bd42a019
PiperOrigin-RevId: 638110795
<!--
If you know who should review your pull request, please assign it to that
person, otherwise the pull request would get assigned randomly.
If your pull request is for a specific language, please add the appropriate
lang label.
-->
Closes#36286
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/36286 from yijiem:enable-win-ee-dns 9276c47326
PiperOrigin-RevId: 635531420
<!--
If you know who should review your pull request, please assign it to that
person, otherwise the pull request would get assigned randomly.
If your pull request is for a specific language, please add the appropriate
lang label.
-->
Closes#36509
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/36509 from ctiller:transport-refs-3 2771a2b0e1
PiperOrigin-RevId: 633240374
This has been stable in CI since May 1st (over a week), zero flakes in EE client-enabled end2end tests.
Closes#36572
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/36572 from drfloob:enable-win-ee-client 8f733316ee
PiperOrigin-RevId: 632233766
The default value of two experiments are set to true because they have already been enabled internally.
Closes#36361
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/36361 from Vignesh2208:experiments-expiry e244a4f30e
PiperOrigin-RevId: 625022979
The pick_first policy creates a list of subchannels for each resolver update and then iterates over the list, attempting to connect to each subchannel in turn, until one of them succeeds. However, once a subchannel does succeed, the policy unrefs the other subchannels but still retains a bunch of now-unnecessary state in the subchannel list itself. This wastes a bunch of memory, especially now that petiole policies are delegating to pick_first. This PR contains a new pick_first implementation that stops retaining that state, which significantly reduces per-channel memory.
There is one behavior change here, which is that if we have a connected subchannel and we get a resolver update that no longer includes that address, we now go IDLE instead of proactively trying to connect to the new addresses.
Closes#34766
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/34766 from markdroth:pick_first_free_memory_after_connecting 7236b4321f
PiperOrigin-RevId: 623887639
It looks like we're settling on moving this functionality into the channel itself - so removing the filter makes sense (especially since it had some significant bugs)
Closes#36213
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/36213 from ctiller:x-ex 3040dcec95
PiperOrigin-RevId: 620996358
Remove rolled out experiments v3_compression_filter, v3_server_auth_filter; Extend WIP v3_channel_idle_filter for a bit, and extend work_serializer_clears_time_cache until OSS issues with work_serializer_dispatch are resolved.
Closes#36166
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/36166 from ctiller:updat3 2c7223f63b
PiperOrigin-RevId: 619057925
<!--
If you know who should review your pull request, please assign it to that
person, otherwise the pull request would get assigned randomly.
If your pull request is for a specific language, please add the appropriate
lang label.
-->
Closes#35992
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/35992 from yousukseung:work-serializer-windows 7b21f46dde
PiperOrigin-RevId: 610042481
<!--
If you know who should review your pull request, please assign it to that
person, otherwise the pull request would get assigned randomly.
If your pull request is for a specific language, please add the appropriate
lang label.
-->
Closes#35573
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/35573 from yijiem:enable-oss-ee-dns-posix-real 1df91d1d84
PiperOrigin-RevId: 609193851
Disabling before the 1.62 branch cut.
There is a flake in one specific interop test: cloud_to_prod_auth:c++:*:oauth2_auth_token:tls, with * being default or gateway_v4. There is also an unresolved PHP crash on test shutdown, just debug builds on Mac (not Linux or Windows).
Closes#35819
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/35819 from drfloob:disable-posix-ee-client c42e8b8c20
PiperOrigin-RevId: 604474559
This reverts commit 6318e9e7e9.
<!--
If you know who should review your pull request, please assign it to that
person, otherwise the pull request would get assigned randomly.
If your pull request is for a specific language, please add the appropriate
lang label.
-->
Closes#35667
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/35667 from ctiller:a 032999b51e
PiperOrigin-RevId: 601495207
<!--
If you know who should review your pull request, please assign it to that
person, otherwise the pull request would get assigned randomly.
If your pull request is for a specific language, please add the appropriate
lang label.
-->
Closes#35573
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/35573 from yijiem:enable-oss-ee-dns-posix-real 017b99312f
PiperOrigin-RevId: 601245249
- `memory_pressure_controller` finally - allows deletion of pid_controller throughout the codebase
- `overload_protection` - one of the http2 rapid reset mitigations
- `red_max_concurrent_streams` - another http2 rapid reset mitigation
Closes#35426
COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/35426 from ctiller:new-years-cleanse 4651672e7e
PiperOrigin-RevId: 595205029
Craig Tiller
Yousuk Seung
Mark D. Roth
Yash Tibrewal
AJ Heller
Yijie Ma
Vignesh Babu
Alisha Nanda
gRPC Team Bot
Tanvi Jagtap