Add a new binary that runs all core end2end tests in fuzzing mode.
In this mode FuzzingEventEngine is substituted for the default event
engine. This means that time is simulated, as is IO. The FEE gets
control of callback delays also.
In our tests the `Step()` function becomes, instead of a single call to
`completion_queue_next`, a series of calls to that function and
`FuzzingEventEngine::Tick`, driving forward the event loop until
progress can be made.
PR guide:
---
**New binaries**
`core_end2end_test_fuzzer` - the new fuzzer itself
`seed_end2end_corpus` - a tool that produces an interesting seed corpus
**Config changes for safe fuzzing**
The implementation tries to use the config fuzzing work we've previously
deployed in api_fuzzer to fuzz across experiments. Since some
experiments are far too experimental to be safe in such fuzzing (and
this will always be the case):
- a new flag is added to experiments to opt-out of this fuzzing
- a new hook is added to the config system to allow variables to
re-write their inputs before setting them during the fuzz
**Event manager/IO changes**
Changes are made to the event engine shims so that tcp_server_posix can
run with a non-FD carrying EventEngine. These are in my mind a bit
clunky, but they work and they're in code that we expect to delete in
the medium term, so I think overall the approach is good.
**Changes to time**
A small tweak is made to fix a bug initializing time for fuzzers in
time.cc - we were previously failing to initialize
`g_process_epoch_cycles`
**Changes to `Crash`**
A version that prints to stdio is added so that we can reliably print a
crash from the fuzzer.
**Changes to CqVerifier**
Hooks are added to allow the top level loop to hook the verification
functions with a function that steps time between CQ polls.
**Changes to end2end fixtures**
State machinery moves from the fixture to the test infra, to keep the
customizations for fuzzing or not in one place. This means that fixtures
are now just client/server factories, which is overall nice.
It did necessitate moving some bespoke machinery into
h2_ssl_cert_test.cc - this file is beginning to be problematic in
borrowing parts but not all of the e2e test machinery. Some future PR
needs to solve this.
A cq arg is added to the Make functions since the cq is now owned by the
test and not the fixture.
**Changes to test registration**
`TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is
used as a first depot for test information.
The gtest version of these tests: queries that registry to manually
register tests with gtest. This ultimately changes the name of our tests
again (I think for the last time) - the new names are shorter and more
readable, so I don't count this as a regression.
The fuzzer version of these tests: constructs a database of fuzzable
tests that it can consult to look up a particular suite/test/config
combination specified by the fuzzer to fuzz against. This gives us a
single fuzzer that can test all 3k-ish fuzzing ready tests and cross
polinate configuration between them.
**Changes to test config**
The zero size registry stuff was causing some problems with the event
engine feature macros, so instead I've removed those and used GTEST_SKIP
in the problematic tests. I think that's the approach we move towards in
the future.
**Which tests are included**
Configs that are compatible - those that do not do fd manipulation
directly (these are incompatible with FuzzingEventEngine), and those
that do not join threads on their shutdown path (as these are
incompatible with our cq wait methodology). Each we can talk about in
the future - fd manipulation would be a significant expansion of
FuzzingEventEngine, and is probably not worth it, however many uses of
background threads now should probably evolve to be EventEngine::Run
calls in the future, and then would be trivially enabled in the fuzzers.
Some tests currently fail in the fuzzing environment, a
`SKIP_IF_FUZZING` macro is used for these few to disable them if in the
fuzzing environment. We'll burn these down in the future.
**Changes to fuzzing_event_engine**
Changes are made to time: an exponential sweep forward is used now -
this catches small time precision things early, but makes decade long
timers (we have them) able to be used right now. In the future we'll
just skip time forward to the next scheduled timer, but that approach
doesn't yet work due to legacy timer system interactions.
Changes to port assignment: we ensure that ports are legal numbers
before assigning them via `grpc_pick_port_or_die`.
A race condition between time checking and io is fixed.
---------
Co-authored-by: ctiller <ctiller@users.noreply.github.com>
Need to do the channelz bit prior to the finishing the op bit.
<!--
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.
-->
Resolve `TESTING_VERSION` to `dev-VERSION` when the job is initiated by
a user, and not the CI. Override this behavior with setting
`FORCE_TESTING_VERSION`.
This solves the problem with the manual job runs executed against a WIP
branch (f.e. a PR) overriding the tag of the CI-built image we use for
daily testing.
The `dev` and `dev-VERSION` "magic" values supported by the
`--testing_version` flag:
- `dev` and `dev-master` and treated as `master`: all
`config.version_gte` checks resolve to `True`.
- `dev-VERSION` is treated as `VERSION`: `dev-v1.55.x` is treated as
simply `v1.55.x`. We do this so that when manually running jobs for old
branches the feature skip check still works, and unsupported tests are
skipped.
This changes will take care of all langs/branches, no backports needed.
ref b/256845629
Currently, we are not very consistent in what we assume the initial
state of an LB policy will be and whether or not we assume that it will
immediately report a new picker when it gets its initial address update;
different parts of our code make different assumptions. This PR
establishes the convention that LB policies will be assumed to start in
state CONNECTING and will *not* be assumed to report a new picker
immediately upon getting their initial address update, and we now assume
that convention everywhere consistently.
This is a preparatory step for changing policies like round_robin to
delegate to pick_first, which I'm working on in #32692. As part of that
change, we need pick_first to not report a connectivity state until it
actually sees the connectivity state of the underlying subchannels, so
that round_robin knows when to swap over to a new child list without
reintroducing the problem fixed in #31939.
To fix this error
```
test/core/security/grpc_authorization_engine_test.cc:88:32: error: unknown type name 'Json'; did you mean 'experimental::Json'?
ParseAuditLoggerConfig(const Json&) override {
^~~~
experimental::Json
```
This check compares the host portion of the target name to the authority
header, but in common use cases (e.g. GCS) they may not coincide.
Additionally, this check does not happen in the Go and Java ALTS stacks.
This makes the JSON API visible as part of the C-core API, but in the
`experimental` namespace. It will be used as part of various
experimental APIs that we will be introducing in the near future, such
as the audit logging API.
WireWriter implementation schedules actions to be run by `ExecCtx`. We
should flush pending actions before destructing
`end2end_testing::g_transaction_processor`, which need to be alive to
handle the scheduled actions. Otherwise,
we get heap-use-after-free error because the testing fixture
(`end2end_testing::g_transaction_processor`) is destructed before all
the scheduled actions are run.
This lowers end2end binder transport test failure rate from 0.23% to
0.15%, according to internal tool that runs the test for 15000 times
under various configuration.
This file does not contain a shebang, and whenever I try and run it it
wedges my console into some weird state.
There's a .sh file with the same name that should be run instead. Remove
the executable bit of the thing we shouldn't run directly so we, like,
don't.
<!--
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.
-->
Previously the error message didn't provide much context, example:
```py
Traceback (most recent call last):
File "/tmpfs/tmp/tmp.BqlenMyXyk/grpc/tools/run_tests/xds_k8s_test_driver/tests/affinity_test.py", line 127, in test_affinity
self.assertLen(
AssertionError: [] has length of 0, expected 1.
```
ref b/279990584.
This PR implements a work-stealing thread pool for use inside
EventEngine implementations. Because of historical risks here, I've
guarded the new implementation behind an experiment flag:
`GRPC_EXPERIMENTS=work_stealing`. Current default behavior is the
original thread pool implementation.
Benchmarks look very promising:
```
bazel test \
--test_timeout=300 \
--config=opt -c opt \
--test_output=streamed \
--test_arg='--benchmark_format=csv' \
--test_arg='--benchmark_min_time=0.15' \
--test_arg='--benchmark_filter=_FanOut' \
--test_arg='--benchmark_repetitions=15' \
--test_arg='--benchmark_report_aggregates_only=true' \
test/cpp/microbenchmarks:bm_thread_pool
```
2023-05-04: `bm_thread_pool` benchmark results on my local machine (64
core ThreadRipper PRO 3995WX, 256GB memory), comparing this PR to
master:
2023-05-04: `bm_thread_pool` benchmark results in the Linux RBE
environment (unsure of machine configuration, likely small), comparing
this PR to master.
---------
Co-authored-by: drfloob <drfloob@users.noreply.github.com>
One TXT lookup query can return multiple TXT records (see the following
example). `EventEngine::DNSResolver` should return all of them to let
the caller (e.g. `event_engine_client_channel_resolver`) decide which
one they would use.
```
$ dig TXT wikipedia.org
; <<>> DiG 9.18.12-1+build1-Debian <<>> TXT wikipedia.org
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 49626
;; flags: qr rd ra; QUERY: 1, ANSWER: 3, AUTHORITY: 0, ADDITIONAL: 1
;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 512
;; QUESTION SECTION:
;wikipedia.org. IN TXT
;; ANSWER SECTION:
wikipedia.org. 600 IN TXT "google-site-verification=AMHkgs-4ViEvIJf5znZle-BSE2EPNFqM1nDJGRyn2qk"
wikipedia.org. 600 IN TXT "yandex-verification: 35c08d23099dc863"
wikipedia.org. 600 IN TXT "v=spf1 include:wikimedia.org ~all"
```
Note that this change also deviates us from the iomgr's DNSResolver API
which uses std::string as the result type.
<!--
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.
-->
<!--
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.
-->
---------
Co-authored-by: Sergii Tkachenko <hi@sergii.org>
Reverts grpc/grpc#32924. This breaks the build again, unfortunately.
From `test/core/event_engine/cf:cf_engine_test`:
```
error: module .../grpc/test/core/event_engine/cf:cf_engine_test does not depend on a module exporting 'grpc/support/port_platform.h'
```
@sampajano I recommend looking into CI tests to catch iOS problems
before merging. We can enable EventEngine experiments in the CI
generally once this PR lands, but this broken test is not one of those
experiments. A normal build should have caught this.
cc @HannahShiSFB
<!--
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.
-->
The job run time was creeping to the 2h timeout. Let's bump it to 3h.
Note that this is `master` branch, so it also includes the build time
every time we commit to grpc/grpc.
ref b/280784903
This PR tries to correct a confusing message. For my little
understanding, it seems that if the code reaches that level, it is not
correct to say that setting TCP_USER_TIMEOUT failed, but actually it
succeeded (setsockopt returned 0) but is different to the previous value
option.
I haven't found the `gpr_log` description, so I hope the PR is correct.
---------
Signed-off-by: Joan Fontanals Martinez <joan.martinez@jina.ai>
### Description
Fix https://github.com/grpc/grpc/issues/24470.
Adding one example which demonstrate the following use cases:
* Generate RPC ID on client side and propagate to server.
* Context propagation from client to server.
* Context propagation between different server interceptors and the
server handler.
## Use:
1. Start server: `python3 -m async_greeter_server_with_interceptor`
2. Start client: `python3 -m async_greeter_client`
### Expected Logs:
* On client side:
```
Sending request with rpc id: 73bb98beff10c2dd7b9f2252a1e2039e
Greeter client received: Hello, you!
```
* On server side:
```
INFO:root:Starting server on [::]:50051
INFO:root:Interceptor1 called with rpc_id: default
INFO:root:Interceptor2 called with rpc_id: Interceptor1-default
INFO:root:Handle rpc with id Interceptor2-Interceptor1-73bb98beff10c2dd7b9f2252a1e2039e in server handler.
```
<!--
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.
-->
Reverts grpc/grpc#33002. Breaks internal builds:
`.../privacy_context:filters does not depend on a module exporting
'.../src/core/lib/channel/context.h'`
Change call attributes to be stored in a `ChunkedVector` instead of
`std::map<>`, so that the storage can be allocated on the arena. This
means that we're now doing a linear search instead of a map lookup, but
the total number of attributes is expected to be low enough that that
should be okay.
Also, we now hide the actual data structure inside of the
`ServiceConfigCallData` object, which required some changes to the
`ConfigSelector` API. Previously, the `ConfigSelector` would return a
`CallConfig` struct, and the client channel would then use the data in
that struct to populate the `ServiceConfigCallData`. This PR changes
that such that the client channel creates the `ServiceConfigCallData`
before invoking the `ConfigSelector`, and it passes the
`ServiceConfigCallData` into the `ConfigSelector` so that the
`ConfigSelector` can populate it directly.
The protection is added at `xds_http_rbac_filter.cc` where we read the
new field. With this disabling the feature, nothing from things like
`xds_audit_logger_registry.cc` shall be invoked.
Makes some awkward fixes to compression filter, call, connected channel
to hold the semantics we have upheld now in tests.
Once the fixes described here
https://github.com/grpc/grpc/blob/master/src/core/lib/channel/connected_channel.cc#L636
are in this gets a lot less ad-hoc, but that's likely going to be
post-landing promises client & server side.
We specifically need special handling for server side cancellation in
response to reads wrt the inproc transport - which doesn't track
cancellation thoroughly enough itself.
<!--
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.
-->
---------
Co-authored-by: ctiller <ctiller@users.noreply.github.com>
<!--
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.
-->
1. `GrpcAuthorizationEngine` creates the logger from the given config in
its ctor.
2. `Evaluate()` invokes audit logging when needed.
---------
Co-authored-by: rockspore <rockspore@users.noreply.github.com>
See `event_engine.h` for the contract change. All other changes are
cleanup.
I confirmed that both the Posix and Windows implementations comply with
this already.
On Windows, the `WindowsEventEngineListener` will only call
`on_shutdown` after all `SinglePortSocketListener`s have been destroyed,
which ensures that no `on_accept` callback will be executed, even if
there is still trailing overlapped activity on the listening socket.
On Posix, the `PosixEngineListenerImpl` will only call `on_shutdown`
after all `AsyncConnectionAcceptor`s have been destroyed, which ensures
`EventHandle::OrphanHandle` has been called. The `OrphanHandle` contract
indicates that all existing notify closures must have already run. The
implementation looks to comply, so if it does not, that's a bug.
3aae08d25e/src/core/lib/event_engine/posix_engine/event_poller.h (L48-L50)
<!--
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.
-->
Whilst the per cpu counters probably help single channel contention, we
think it's likely that they're a pessimization when taken fleetwide.
<!--
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.
-->
Spin off from https://github.com/grpc/grpc/pull/32701.
<!--
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.
-->
Spin off from https://github.com/grpc/grpc/pull/32701.
<!--
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.
-->
Craig Tiller
Hannah Shi
molto
Yingwei Fan
Mark D. Roth
Sergii Tkachenko
AJ Heller
Benjamin Peterson
Esun Kim
Matthew Stevenson
Eugene Ostroukhov
Ming-Chuan
Yijie Ma
Xuan Wang
Easwar Swaminathan
Bulat Gaifullin
Joan Fontanals
Luwei Ge
Vignesh Babu
