[WIP] GCS client test failure flakiness#34656
Merged
pcmoritz merged 2 commits intoray-project:masterfrom Apr 22, 2023
Merged
Conversation
fishbone
approved these changes
Apr 21, 2023
Contributor
|
I'm merging this now since I was debugging a PR that ran into the same issue I think, seeing if this fixes it :) |
ProjectsByJackHe
pushed a commit
to ProjectsByJackHe/ray
that referenced
this pull request
May 4, 2023
Why are these changes needed? Right now the theory is as follow. pubsub io service is created and run inside the GcsServer. That means if pubsub io service is accessed after GCSServer GC'ed, it will segfault. Right now, upon teardown, when we call rpc::DrainAndResetExecutor, this will recreate the Executor thread pool. Upon teardown, If DrainAndResetExecutor -> GcsServer's internal pubsub posts new SendReply to the newly created threadpool -> GcsServer.reset -> pubsub io service GC'ed -> SendReply invoked from the newly created thread pool, it will segfault. NOTE: the segfault is from pubsub service if you see the failure ray-project#2 0x7f92034d9129 in ray::rpc::ServerCallImpl<ray::rpc::InternalPubSubGcsServiceHandler, ray::rpc::GcsSubscriberPollRequest, ray::rpc::GcsSubscriberPollReply>::HandleRequestImpl()::'lambda'(ray::Status, std::__1::function<void ()>, std::__1::function<void ()>)::operator()(ray::Status, std::__1::function<void ()>, std::__1::function<void ()>) const::'lambda'()::operator()() const /proc/self/cwd/bazel-out/k8-opt/bin/_virtual_includes/grpc_common_lib/ray/rpc/server_call.h:212:48 As a fix, I only drain the thread pool. And then reset it after all operations are fully cleaned up (only from tests). I think there's no need to reset for regular proc termination like raylet, gcs, core workers. Related issue number Closes ray-project#34344 Signed-off-by: SangBin Cho <rkooo567@gmail.com> Signed-off-by: Jack He <jackhe2345@gmail.com>
architkulkarni
pushed a commit
to architkulkarni/ray
that referenced
this pull request
May 16, 2023
Why are these changes needed? Right now the theory is as follow. pubsub io service is created and run inside the GcsServer. That means if pubsub io service is accessed after GCSServer GC'ed, it will segfault. Right now, upon teardown, when we call rpc::DrainAndResetExecutor, this will recreate the Executor thread pool. Upon teardown, If DrainAndResetExecutor -> GcsServer's internal pubsub posts new SendReply to the newly created threadpool -> GcsServer.reset -> pubsub io service GC'ed -> SendReply invoked from the newly created thread pool, it will segfault. NOTE: the segfault is from pubsub service if you see the failure #2 0x7f92034d9129 in ray::rpc::ServerCallImpl<ray::rpc::InternalPubSubGcsServiceHandler, ray::rpc::GcsSubscriberPollRequest, ray::rpc::GcsSubscriberPollReply>::HandleRequestImpl()::'lambda'(ray::Status, std::__1::function<void ()>, std::__1::function<void ()>)::operator()(ray::Status, std::__1::function<void ()>, std::__1::function<void ()>) const::'lambda'()::operator()() const /proc/self/cwd/bazel-out/k8-opt/bin/_virtual_includes/grpc_common_lib/ray/rpc/server_call.h:212:48 As a fix, I only drain the thread pool. And then reset it after all operations are fully cleaned up (only from tests). I think there's no need to reset for regular proc termination like raylet, gcs, core workers. Related issue number Closes ray-project#34344 Signed-off-by: SangBin Cho <rkooo567@gmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why are these changes needed?
Right now the theory is as follow.
NOTE: the segfault is from pubsub service if you see the failure
As a fix, I only drain the thread pool. And then reset it after all operations are fully cleaned up (only from tests). I think there's no need to reset for regular proc termination like raylet, gcs, core workers.
Related issue number
Closes #34344
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.