fix: Implement timeout layer correctly by using timeout #4059
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Outdated suggestions cannot be applied.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.
TimeoutLayer is using
tokio::time::timeout
to implement timeout for operations. And IO Operations insidesreader
,writer
will usePin<Box<tokio::time::Sleep>>
to track the timeout.This might introduce a bit overhead for IO operations, but it's the only way to implement timeout correctly. We used to implement timeout layer in zero cost way that only stores a
std::time::Instant
and check the timeout by comparing the instant with current time. However, it doesn't works for all cases.For examples, users TCP connection could be in Busy ESTAB state. In this state, no IO event will be emit. The runtime will never poll our future again. From the application side, this future is hanging forever until this TCP connection is closed for reaching the linux net.ipv4.tcp_retries2 times.