We have a bunch of experiments testing against core e2e - and this is
good for robustness, bad for CI times.
We also have a bunch of marginal but overall necessary fixtures in the
e2e suites - again good for robustness, bad for CI times.
We can eliminate some of the cross product though, and I think safely:
run experiments on a broad range of suites, but not *ALL* the suites,
and get a bunch of our CI time back.
Here I introduce an environment variable: `GRPC_CI_EXPERIMENTS` that's
set when running bazel @experiment= configs, cleared otherwise (so we
can still execute those tests directly when necessary). When that env
var is set we filter out a bunch of suites from the test configurations.
Add some basic metrics to work serializer, keep them process wide for
now (though it may be interesting to get these into channelz in the
future).
Collected are:
- time spent running a work serializer when it starts
- time spent actually executing work when the work serializer runs
- number of items executed each run
A high disparity between the first two indicates our dispatching
mechanism is adding large amounts of latency (perhaps due to thread
starvation like effects).
A high value for any of these indicate contention on the serializer.
It's likely a future iteration on these will select different metrics -
I'm not *entirely* sure which will be useful in production analysis yet.
I'm using `std::chrono::steady_clock` here for precision (nanoseconds)
with a compact representation (better than timespec) and a robust &
portable api - I think it's appropriate for metrics, but wouldn't use it
much beyond that at this point.
This has been stable for a bit, everywhere that the EventEngine is
enabled. Going forward, I think the event_engine_{client|listener}
experiments can probably be used to regulate thread-pool-specific
issues.
---------
Co-authored-by: drfloob <drfloob@users.noreply.github.com>
Move the SSL_CTX to the level of the credentials rather than the
subchannel.
The SSL_CTX should only get created once per credential rather than once
per subchannel.
We should observe no behavior change with this PR, only efficiency
gains.
Most recent attempt was #34320, reverted in #34335.
The first commit here is a pure revert. The second commit fixes the
outlier_detection unit test to pass both with and without the
experiment.
To fix the following build error with the head of abseil
```
/var/local/git/grpc/test/core/tsi/ssl_transport_security_utils_test.cc:231:42: error: no member named 'StrCat' in namespace 'absl'
return absl::InternalError(absl::StrCat("Client error:", client_err));
~~~~~~^
/var/local/git/grpc/test/core/tsi/ssl_transport_security_utils_test.cc:238:42: error: no member named 'StrCat' in namespace 'absl'
return absl::InternalError(absl::StrCat("Server error:", server_err));
~~~~~~^
```
The previous approach of generating strings was not converging well.
Instead, load a bitfield from the protobuf and use the bits to select
experiments. The fuzzers can explore this space swiftly.
Downside is that as experiments rotate in/out the corpus gets a bit
messed up, but I'm reasonably confident we'll recover quickly.
---------
Co-authored-by: ctiller <ctiller@users.noreply.github.com>
Splitting off from https://github.com/grpc/grpc/pull/34273
<!--
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.
-->
In certain situations the current flow control algorithm can result in
sending one flow control update write for every write sent (known
situation: rollout of promise based server calls with qps_test).
Fix things up so that the updates are only sent when truly needed, and
then fix the fallout (turns out our fuzzer had some bugs)
I've placed actual logic changes behind an experiment so that it can be
incrementally & safely rolled out.
This should address one of the failures we're seeing in #34224.
The test failure is caused by the changes in timing triggering a race
condition. In the code at head, we delay sending out the subscription
for the first CDS watch until we've already seen the other two CDS
watches, because the previous send_message op has not yet completed, and
by the time it does, we've seen all 3 watches, so we can send a
subscription for all 3 at the same time. With the WorkSerializer change,
the send_message op is complete by the time we see the first CDS watch,
so we subscribe to only that resource, and then later add the other two.
The result is that we'll NACK twice with two different messages, the
first one including only the error about the first resource, and the
second one including all three.
I suspect this same race condition would have been triggered eventually
by the EventEngine migration anyway; the current test basically depends
on the single-thread timing of the iomgr approach. So I'm addressing it
by replacing the e2e test with a unit test that covers the same cases
without the timing issue.
Rolls forward part of the dualstack changes, mostly from #33427 and a
little bit from #32692, both of which were reverted in #33718.
Specifically:
- For petiole policies, unconditionally start health watch on
subchannels, even if client side health checking is not enabled; in this
case, the health watch will report the subchannel's raw connectivity
state.
- Fix edge cases in health check reporting that occur when a watcher is
started before the initial state is reported.
- When client-side health checking fails, add the subchannel's address
to the RPC failure status message.
- Outlier detection now works only via the health checking watch, not
via the raw connectivity state watch.
- Remove now-unnecessary hack to ensure that outlier detection does not
work for pick_first.
This rolls forward only the pick_first changes from #32692, which were
rolled back in #33718. Specifically:
- Changes PF to use its own subchannel list implementation instead of
using the subchannel_list library, since the latter will be going away
with the dualstack changes.
- As a result of no longer using the subchannel_list library, PF no
longer needs to set the `GRPC_ARG_INHIBIT_HEALTH_CHECKING` channel arg.
- Adds an option to start a health watch on the chosen subchannel, to be
used in the future when pick_first is the child of a petiole policy.
(Currently, this code is not actually called anywhere.)
<!--
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: Mark D. Roth <roth@google.com>
Co-authored-by: markdroth <markdroth@users.noreply.github.com>
Co-authored-by: ctiller <ctiller@users.noreply.github.com>
Pipe-like type (has a send end, a receive end, and a closing mechanism)
for cross-activity transfers.
---------
Co-authored-by: ctiller <ctiller@users.noreply.github.com>
If we get a readable event on an fd and both the following happens:
- c-ares does *not* read all bytes off the fd
- c-ares removes the fd from the set ARES_GETSOCK_READABLE
... then we have a busy loop here, where we'd keep asking c-ares to
process an fd that it no longer cares about.
This is indirectly related to a change in this code one month ago:
https://github.com/grpc/grpc/pull/33942 - before that change, c-ares
would close the socket when it called
[handle_error](7f3262312f/src/lib/ares_process.c (L707))
and so `IsFdStillReadableLocked` would start returning `false`, causing
us to get away with [this
loop](f6a994229e/src/core/ext/filters/client_channel/resolver/dns/c_ares/grpc_ares_wrapper.cc (L371)).
Now, because `IsFdStillReadableLocked` will keep returning true (because
of our overridden `close` API), we'll loop forever.
Building out a new framing layer for chttp2.
The central idea here is to have the framing layer be solely responsible
for serialization of frames, and their deserialization - the framing
layer can reject frames that have invalid syntax - but the enacting of
what that frame means is left to a higher layer.
This class will become foundational for the promise conversion of chttp2
- by eliminating action from the parsing of frames we can reuse this
sensitive code.
Right now the new layer is inactive - there's a test that exercises it
relatively well, and not much more. In the next PRs I'll add an
experiments to enable using this layer or the existing code in the
writing and reading paths.
---------
Co-authored-by: ctiller <ctiller@users.noreply.github.com>
This is the initial implementation of the chaotic-good client transport
write path. There will be a follow-up PR to fulfill the read path.
<!--
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.
-->
We enabled OpenSSL3 testing with #31256 and missed a failing test
It wasn't running before, so this isn't a regression - disabling it so
master doesn't fail while we figure out how to fix it.
Update from gtcooke94:
This PR adds support to build gRPC and it's tests with OpenSSL3. There were some
hiccups with tests as the tests with openssl haven't been built or exercised in a
few months, so they needed some work to fix.
Right now I expect all test files to pass except the following:
- h2_ssl_cert_test
- ssl_transport_security_utils_test
I confirmed locally that these tests fail with OpenSSL 1.1.1 as well,
thus we are at least not introducing regressions. Thus, I've added compiler directives around these tests so they only build when using BoringSSL.
---------
Co-authored-by: Gregory Cooke <gregorycooke@google.com>
Co-authored-by: Esun Kim <veblush@google.com>
- add debug-only `WorkSerializer::IsRunningInWorkSerializer()` method
and use it in client_channel to verify that subchannels are destroyed in
the `WorkSerializer`
- note: this mechanism uses `std:🧵:id`, so I had to exclude
work_serializer.cc from the core_banned_constructs check
- fix `WorkSerializer::Run()` to unref the callback before releasing
ownership of the `WorkSerializer`, so that any refs captured by the
`std::function<>` will be released before releasing ownership
- fix the WRR timer callback to hop into the `WorkSerializer` to release
its ref to the picker, since that transitively releases refs to
subchannels
- fix subchannel connectivity state notifications to unref the watcher
inside the `WorkSerializer`, since the watcher often transitively holds
refs to subchannels
Proposed alternative to https://github.com/grpc/grpc/pull/34024.
This version has a simpler, faster busy-count implementation based on a
sharded set of atomic counts: fast increment/decrement operations,
relatively slower summation of total counts (which need to happen much
less frequently).
Craig Tiller
Mark D. Roth
AJ Heller
Esun Kim
nanahpang
Gregory Cooke
Romain Geissler @ Amadeus
Eugene Ostroukhov
Yash Tibrewal
apolcyn
jrandolf