Convert BytesTransportResponse when proxying response from/to local node #135873
Conversation
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
execution, to a BytesTransportResponse as opposed to materializing the response object on heap. When a proxy node acts as a proxy to query its local data, and the coordinating node is on a different version than the proxy node, the response will fail to deserialize in the coord node because it was written with the version of the proxy node as opposed to that of the coord (target) node. This is because DirectResponseChannel does not read and write such response, which would lead to it being converted to the right format. This commit attempts to fix this problem by tracking the version used to write the response, and conditionally converting it in the ProxyRequestHandler.
javanna
added
>bug
:Distributed Coordination/Network
|
Hi @javanna, I've created a changelog YAML for you. |
elasticsearchmachine
added
the
Team:Distributed Coordination
|
Pinging @elastic/es-distributed-coordination (Team:Distributed Coordination) |
DaveCTurner
reviewed
Oct 3, 2025
server/src/main/java/org/elasticsearch/transport/TransportActionProxy.java
Outdated
Show resolved
Hide resolved
…onProxy.java Co-authored-by: David Turner <[email protected]>
javanna
commented
Oct 3, 2025
| try { | ||
| channel.sendResponse(convertedResponse); | ||
| } finally { | ||
| convertedResponse.decRef(); |
There was a problem hiding this comment.
I'd initially forgotten this, can you double check that this is what I need to be doing please?
javanna
commented
Oct 3, 2025
| convertedResponse.decRef(); | ||
| } | ||
| } catch (IOException e) { | ||
| throw new UncheckedIOException(e); |
There was a problem hiding this comment.
should I throw or send the exception as a response? I am assuming that exceptions thrown are already caught and sent as responses, but good to double check.
javanna
commented
Oct 3, 2025
| public void writeTo(StreamOutput out) throws IOException { | ||
| out.writeString(targetNode); | ||
| if (out.getTransportVersion().supports(transportVersion1)) { | ||
| out.writeBoolean(true); |
There was a problem hiding this comment.
This reproduces the problem in the tests I added, which is fixed with the conversion this change introduces.
javanna
commented
Oct 3, 2025
| "internal:test", | ||
| cancellable, | ||
| // For a proxy node proxying to itself, the response is sent directly, without it being read by the proxy layer | ||
| r -> { throw new AssertionError(); }, |
There was a problem hiding this comment.
this diff is misleading! I only changed this line in testSendLocalRequest
|
Thanks for the suggestion, I added tests @DaveCTurner. |
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.
#127112 introduced
BytesTransportResponseto be used in search batched execution, so thatNodeQueryResponsecould be written asBytesTransportResponseas opposed to materializing the response object on heap.When a proxy node acts as a proxy to query its local data, and the coordinating node is on a different version than the proxy node, the response will fail to deserialize in the coord node because it was written with the version of the proxy node as opposed to that of the coord (target) node. This is because
DirectResponseChannelskips the step of reading and writing back such response, which would lead to it being converted to the right format.This commit attempts to fix this problem by tracking the version used to write the response, and conditionally converting it in the ProxyRequestHandler.
There may be better ways to fix this problem, let's use this draft to align on the problem and agree on possible solutions. I am happy to iterate on it, as this currently blocks merging #135549 due to the serialization change it introduces.
Note that search batched executed is still under a feature flag, and disabled by default, so this bug does not currently affect our users.