feat(gw): error propagation #339

ardatan · 2025-08-14T15:32:34Z

Rewrite of #299 within the new structure

Normalize error paths with _entities into the real paths using flatten paths
Bring back some of old test components so I could write the unit tests based on local schemas without actual HTTP requests
Add serviceName and code to the subgraph errors while passing it to the final gw response
Remove traverse_and_callback_mut as it is no longer needed because we collect indexes in paths and use it with flatten path to get the mutable reference in the final response without traversing again

gemini-code-assist

Summary of Changes

Hello @ardatan, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request significantly overhauls the error propagation mechanism within the GraphQL gateway, focusing on providing more precise and actionable error information to clients. The changes involve restructuring how errors are represented internally, introducing a sophisticated normalization layer to correctly map subgraph errors to the gateway's response paths, and enhancing data traversal to support this mapping. These improvements ensure that even in complex federated query scenarios, errors are clearly attributed and located, greatly aiding debugging and client-side error handling. The refactoring also includes internal naming clarifications and adoption of standard asynchronous primitives.

Highlights

Enhanced Error Structure: The core error handling mechanism has been significantly upgraded. GraphQLError now supports structured extensions using a HashMap, enabling richer metadata like serviceName and specific error codes to be attached to errors. This provides clearer context for debugging and client-side error handling.
Intelligent Error Path Normalization: A new error_normalization module has been introduced to intelligently map errors originating from subgraphs (especially in federated queries where entities are flattened) back to their logical paths in the gateway's final response. This ensures that clients receive accurate error locations, even for complex nested data structures.
Improved Data Traversal for Error Mapping: The internal data traversal logic (traverse_and_callback) has been refined to track precise paths, including array indices, using VecDeque. This detailed path information is critical for the new error normalization process, allowing the gateway to pinpoint the exact location of an error within a deeply nested response.
Robust Testing for Error Propagation: The pull request includes comprehensive new test infrastructure, integrating async-graphql for easy subgraph simulation and dedicated fixtures. A new error_propagation test case thoroughly validates the end-to-end error handling, ensuring that errors are correctly captured, normalized, and propagated to the client.
Internal Refactorings and Simplifications: Several key internal components, such as ExecutionContext (now QueryPlanExecutionContext) and HttpExecutionRequest (now SubgraphExecutionRequest), have been renamed for better clarity and consistency. The custom concurrency management has also been replaced with FuturesUnordered for streamlined asynchronous operations.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in issue comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

This pull request significantly refactors error propagation within the GraphQL executor, introducing more structured handling, especially for federated queries. The changes include renaming several components for clarity, improving parallel execution logic, and adding comprehensive tests for error propagation scenarios. I've identified a critical bug where some subgraph errors were being dropped and an inconsistency in how error extensions are applied. Addressing these points will make the error handling much more robust. Overall, this is a great improvement.

lib/executor/src/execution/plan.rs

lib/executor/src/response/error_normalization.rs

github-actions · 2025-08-14T15:34:56Z

Federation Audit Results

189 tests ±0 189 ✅ ±0 3s ⏱️ ±0s
42 suites ±0 0 💤 ±0
42 files ±0 0 ❌ ±0

Results for commit 7d567a0. ± Comparison against base commit 02bf99e.

♻️ This comment has been updated with latest results.

github-actions · 2025-08-14T15:36:25Z

✅ `k6-benchmark` results

     ✓ response code was 200
     ✓ no graphql errors
     ✓ valid response structure

     █ setup

     checks.........................: 100.00% ✓ 42270      ✗ 0    
     data_received..................: 1.2 GB  41 MB/s
     data_sent......................: 17 MB   550 kB/s
     http_req_blocked...............: avg=41.52µs  min=791ns   med=2.09µs   max=19.44ms  p(90)=2.97µs   p(95)=3.43µs  
     http_req_connecting............: avg=36.99µs  min=0s      med=0s       max=19.42ms  p(90)=0s       p(95)=0s      
     http_req_duration..............: avg=104.9ms  min=2.2ms   med=102.9ms  max=298.38ms p(90)=130.36ms p(95)=141.69ms
       { expected_response:true }...: avg=104.9ms  min=2.2ms   med=102.9ms  max=298.38ms p(90)=130.36ms p(95)=141.69ms
     http_req_failed................: 0.00%   ✓ 0          ✗ 14110
     http_req_receiving.............: avg=454.15µs min=31.02µs med=53.45µs  max=90.27ms  p(90)=1.16ms   p(95)=2.8ms   
     http_req_sending...............: avg=101.89µs min=6.55µs  med=12.73µs  max=17.06ms  p(90)=19.06µs  p(95)=30.18µs 
     http_req_tls_handshaking.......: avg=0s       min=0s      med=0s       max=0s       p(90)=0s       p(95)=0s      
     http_req_waiting...............: avg=104.35ms min=2.14ms  med=102.44ms max=256.09ms p(90)=129.63ms p(95)=141.23ms
     http_reqs......................: 14110   468.601621/s
     iteration_duration.............: avg=106.57ms min=16.21ms med=103.33ms max=632.96ms p(90)=130.96ms p(95)=142.43ms
     iterations.....................: 14090   467.937409/s
     vus............................: 50      min=50       max=50 
     vus_max........................: 50      min=50       max=50

kamilkisiela · 2025-08-19T14:58:04Z

lib/executor/src/execution/plan.rs

+                            if let Some(target_paths) =
+                                job.filtered_representations_hashes.get_mut(hash)
+                            {
+                                for indexes_in_path in target_paths {


In traverse_and_callback_mut we give it a path and once it reaches the final segment, it executes the callback.
The traverse of final_response starts at the root, then it moves to the first path segment until it hits the final segment.

In your code, we always start from root, for every element in the target_paths vector.
At least from what I see, but maybe I'm missing something.

What was wrong with traverse_and_callback_mut?

I actually didn't touch the logic. But instead I hold the array indices (the actual values of @ in the flatten path) so I can use them for fixing the error paths in the final response. Since we already know the path, traverse_and_callback_mut is no longer needed because we already use this information to access the entity in the fina lresponse.
Nothing is wrong with it but mutable version of traverse_and_callback is no longer needed. Immutable version is still used to collect the entities and also indices in the paths.

What about the loop within a loop. We start from the final_response for every for indexes_in_path in target_paths

kamilkisiela · 2025-08-19T14:59:49Z

lib/executor/src/execution/plan.rs

    fetch_node_id: i64,
    representation_hashes: Vec<u64>,
-    representation_hash_to_index: HashMap<u64, usize>,
+    filtered_representations_hashes: HashMap<u64, Vec<VecDeque<usize>>>,


Can you add a comment explaining the HashMap value? Or create a dedicated struct if it will make it more obvious?

They are the indexes of the representations. Indexes are the actual values of @ in the flatten path.

kamilkisiela · 2025-08-19T15:00:28Z

lib/executor/src/execution/plan.rs

-                                    job.representation_hash_to_index.get(&hash)
-                                {
-                                    if let Some(entity) = entities.get(*entity_index) {
-                                        // SAFETY: `new_val` is a clone of an entity that lives for `'a`.


This comment is missing, it would be nice to include it, so we know why we use transmute.

kamilkisiela · 2025-08-19T15:02:43Z

lib/executor/src/response/error_normalization.rs

+    if let Some(path_in_error) = &error.path {
+        if let Some(GraphQLErrorPathSegment::String(first_path)) = path_in_error.first() {
+            if first_path == "_entities" {
+                if let Some(GraphQLErrorPathSegment::Index(entity_index)) = path_in_error.get(1) {


Can we somehow reduce the depth level here?

I'll try 👍

gemini-code-assist bot reviewed Aug 14, 2025

View reviewed changes

lib/executor/src/execution/plan.rs Show resolved Hide resolved

lib/executor/src/response/error_normalization.rs Outdated Show resolved Hide resolved

This comment was marked as resolved.

Sign in to view

ardatan marked this pull request as draft August 14, 2025 15:40

ardatan requested review from dotansimha and kamilkisiela August 14, 2025 16:08

ardatan marked this pull request as ready for review August 14, 2025 16:08

ardatan added 5 commits August 17, 2025 17:49

Error Propagation

bbc98cc

Fixes

99d94f8

Clippy

59da743

..

b353ab1

Better

13c3a6e

ardatan force-pushed the error-propagation-v2 branch from d62ddbf to 13c3a6e Compare August 17, 2025 14:49

dotansimha mentioned this pull request Aug 17, 2025

feat(executor, gw): error propagation #299

Closed

dotansimha changed the title ~~Error Propagation~~ feat(gw): error propagation Aug 17, 2025

kamilkisiela reviewed Aug 19, 2025

View reviewed changes

dotansimha force-pushed the main branch 9 times, most recently from 3066ce0 to 53e8bd5 Compare September 1, 2025 12:13

dotansimha force-pushed the main branch 3 times, most recently from e9864f4 to a80cb5b Compare September 1, 2025 13:55

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

feat(gw): error propagation #339

feat(gw): error propagation #339

Uh oh!

ardatan commented Aug 14, 2025 •

edited

Loading

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

Uh oh!

Uh oh!

github-actions bot commented Aug 14, 2025 •

edited

Loading

Uh oh!

This comment was marked as resolved.

github-actions bot commented Aug 14, 2025 •

edited

Loading

Uh oh!

kamilkisiela Aug 19, 2025

Uh oh!

ardatan Aug 19, 2025 •

edited

Loading

Uh oh!

kamilkisiela Aug 21, 2025

Uh oh!

kamilkisiela Aug 19, 2025

Uh oh!

ardatan Aug 19, 2025

Uh oh!

kamilkisiela Aug 19, 2025

Uh oh!

kamilkisiela Aug 19, 2025

Uh oh!

ardatan Aug 19, 2025

Uh oh!

Uh oh!

feat(gw): error propagation #339

Are you sure you want to change the base?

feat(gw): error propagation #339

Uh oh!

Conversation

ardatan commented Aug 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

github-actions bot commented Aug 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Federation Audit Results

Uh oh!

This comment was marked as resolved.

github-actions bot commented Aug 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

✅ k6-benchmark results

Uh oh!

kamilkisiela Aug 19, 2025

Choose a reason for hiding this comment

Uh oh!

ardatan Aug 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kamilkisiela Aug 21, 2025

Choose a reason for hiding this comment

Uh oh!

kamilkisiela Aug 19, 2025

Choose a reason for hiding this comment

Uh oh!

ardatan Aug 19, 2025

Choose a reason for hiding this comment

Uh oh!

kamilkisiela Aug 19, 2025

Choose a reason for hiding this comment

Uh oh!

kamilkisiela Aug 19, 2025

Choose a reason for hiding this comment

Uh oh!

ardatan Aug 19, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

ardatan commented Aug 14, 2025 •

edited

Loading

github-actions bot commented Aug 14, 2025 •

edited

Loading

github-actions bot commented Aug 14, 2025 •

edited

Loading

✅ `k6-benchmark` results

ardatan Aug 19, 2025 •

edited

Loading