4.x Http2Connection should ignore a max concurrent streams setting of zero from the client #7346
Conversation
This file contains 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
…o from the client
oracle-contributor-agreement
bot
added
the
OCA Verified
thegridman
changed the title
danielkec
requested changes
Aug 10, 2023
There was a problem hiding this comment.
This looks more like a bug in the mentioned netty based client, citing the spec:
A value of 0 for SETTINGS_MAX_CONCURRENT_STREAMS SHOULD NOT be treated as special by endpoints. A zero value does prevent the creation of new streams; however, this can also happen for any limit that is exhausted with active streams. Servers SHOULD only set a zero value for short durations; if a server does not wish to accept requests, closing the connection is more appropriate.
Where A zero value does prevent the creation of new streams is important, we have to honor the 0 value like any other.
|
According to my understanding:
|
tomas-langer
previously approved these changes
Aug 11, 2023
danielkec
reviewed
Aug 11, 2023
...grpc/server/src/main/java/io/helidon/nima/tests/integration/grpc/webserver/EventService.java
Outdated
Show resolved
Hide resolved
danielkec
reviewed
Aug 11, 2023
Comment on lines
-246
to
-261
| // Set server MAX_CONCURRENT_STREAMS limit when client sends number lower than hard limit | ||
| // from configuration. Refuse settings if client sends larger number than is configured. | ||
| this.clientSettings.presentValue(Http2Setting.MAX_CONCURRENT_STREAMS) | ||
| .ifPresent(it -> { | ||
| if (http2Config.maxConcurrentStreams() >= it) { | ||
| maxClientConcurrentStreams = it; | ||
| } else { | ||
| Http2GoAway frame = | ||
| new Http2GoAway(0, | ||
| Http2ErrorCode.PROTOCOL, | ||
| "Value of maximum concurrent streams limit " + it | ||
| + " exceeded hard limit value " | ||
| + http2Config.maxConcurrentStreams()); | ||
| connectionWriter.write(frame.toFrameData(clientSettings, 0, Http2Flag.NoFlags.create())); | ||
| } | ||
| }); |
…to explain what it does.
|
Although Daniel has approved this I do not have permissions to merge it so someone else will need to do it. Also, the build fails but I don't think it is related to this change as it is a different test failure every time I re-run the failing validate build. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
The Http2Connection updates some of its configuration based on that sent by the client. One of the config items updated is the "max concurrent streams" setting. The problem is that if the client sends zero for this setting then things can break. I discovered this when a number of our Coherence gRPC tests broke when I switched the server to use Nima and the client is still using grpc-java and Netty.
Client libraries such as
grpc-javahard code this value to be zero and this breaks gRPC because the server cannot send messages down multiple streams. I have added an example of this where I have an existing bi-di channel request open and then send another request from the client that causes the server to send a message down the bi-di channel. This will break without the fix in this PR.The fix I have done here is to ignore changing the "max concurrent streams" setting on the server if the client sends zero. then the grpc-java clients work. It might well be that I have mis-understood something about "max concurrent streams" and how the server should handle a value of zero. Trying to find out anything on the internet was all a bit vague.
I added a test
nima/tests/integration/grpc/server/src/test/java/io/helidon/nima/tests/integration/grpc/webserver/EventsTest.javawhich works with my fix and fails without it.