xds: refactor xdsresource.Endpoint to embed resolver.Endpoint (gRFC A81) #8750

Pranjali-2501 · 2025-12-06T21:16:00Z

This PR updates the internal xdsresource.Endpoint struct to contain a resolver.Endpoint instead of a []string to store the list of addresses associated with the endpoint gRFC A81. This change standardizes how backend information is stored and ensures that attributes (such as Hostname) are correctly associated with the endpoint hierarchy.

Key Changes:

Struct Update:

xdsresource.Endpoint now includes a ResolverEndpoint field (of type resolver.Endpoint) to store addresses and attributes. Remove the existing Address field (of type []string) and store address as a resolver.Endpoint field.

Attribute Handling:

Added SetHostname and GetHostname helpers to manage hostname metadata within resolver.Endpoint.Attributes.

Parsing Logic:

Updated parseEndpoints in unmarshal_eds.go to correctly populate the resolver.Endpoint object.

RELEASE NOTES: N/A

codecov · 2025-12-06T21:18:55Z

Codecov Report

❌ Patch coverage is 85.00000% with 3 lines in your changes missing coverage. Please review.
✅ Project coverage is 83.22%. Comparing base (4c27cc8) to head (578caf3).
⚠️ Report is 4 commits behind head on master.

Files with missing lines	Patch %	Lines
...nternal/xds/xdsclient/xdsresource/unmarshal_eds.go	82.35%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8750      +/-   ##
==========================================
- Coverage   83.28%   83.22%   -0.07%     
==========================================
  Files         418      418              
  Lines       32367    32396      +29     
==========================================
+ Hits        26958    26961       +3     
- Misses       4034     4052      +18     
- Partials     1375     1383       +8

Files with missing lines	Coverage Δ
...rnal/xds/balancer/clusterresolver/configbuilder.go	`94.11% <100.00%> (ø)`
...nternal/xds/xdsclient/xdsresource/unmarshal_eds.go	`92.76% <82.35%> (-1.57%)`	⬇️

... and 26 files with indirect coverage changes

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

easwars · 2025-12-08T19:40:26Z

internal/xds/xdsclient/tests/loadreport_test.go

 // TestConcurrentChannels verifies that we can create multiple gRPC channels
 // concurrently with a shared XDSClient, each of which will create a new LRS
 // stream without any race.
-func (s) TestConcurrentChannels(t *testing.T) {


Please don't remove the (s) receiver.

easwars · 2025-12-08T19:42:15Z

internal/xds/xdsclient/xdsresource/type_eds.go

 	EndpointHealthStatusDegraded
 )

 // Endpoint contains information of an endpoint.


While you are here, could you also please file an issue and add a TODO here to capture the fact that we could completely get rid of this struct, and encode all the information associated with the endpoint in the resolver.Endpoint struct.

easwars · 2025-12-08T19:43:33Z

internal/xds/xdsclient/xdsresource/unmarshal_eds.go

+type hostnameType string
+
+// hostnameKey is the key to store the hostname key attribute in
+// a resolver.Endpoint attribute.
+const hostnameKey = hostnameType("grpc.resolver.xdsresource.hostname")


Instead of a type and a const, this could be simplified as:

type hostnameKeyType struct{}

easwars · 2025-12-08T19:46:16Z

internal/xds/xdsclient/xdsresource/unmarshal_eds.go

+// a resolver.Endpoint attribute.
+const hostnameKey = hostnameType("grpc.resolver.xdsresource.hostname")
+
+// SetHostname sets the hostname key for this resolver.Endpoint attribute.


How about:

// SetHostname returns a copy of the given endpoint with hostname added as an attribute.

easwars · 2025-12-08T19:46:58Z

internal/xds/xdsclient/xdsresource/unmarshal_eds.go

+	if hostname == "" {
+		return endpoint
+	}


Why would someone call this function with an empty hostname? And why is it important to not include the empty hostname in the attributes. If there are valid reasons for doing this, it should be mentioned in the docstring.

The hostname should be non-empty for :authority rewriting.
Added a comment.

easwars · 2025-12-08T19:53:07Z

internal/xds/xdsclient/xdsresource/unmarshal_eds.go

+		endpoint := resolver.Endpoint{
+			Addresses: address,
+		}


Nit: Literal structs with a single field could be initialized on a single line.

easwars · 2025-12-08T19:53:14Z

internal/xds/xdsclient/xdsresource/unmarshal_eds_test.go

+	resolverEndpoint := resolver.Endpoint{
+		Addresses: address,
+	}


Nit: Literal structs with a single field could be initialized on a single line.

easwars · 2025-12-08T19:54:05Z

internal/xds/xdsclient/xdsresource/unmarshal_eds_test.go

 	"google.golang.org/protobuf/types/known/wrapperspb"
 )

+func getResolverEndpoint(addr []string, hostname string) resolver.Endpoint {


The get prefix here makes it look like this is a getter. Please rename it to buildResolverEndpoint or makeResolverEndpoint.

easwars · 2025-12-08T20:18:33Z

internal/xds/balancer/clusterresolver/configbuilder_test.go

+						ResolverEndpoint: resolver.Endpoint{
+							Addresses: []resolver.Address{{Addr: "addr-1-1"}},
+						},


Here and elsewhere in this test file, you could make such lines be one a one-liner since only a single field is being set in the resolver.Endpoint.

easwars · 2025-12-08T20:20:35Z

internal/xds/balancer/clusterresolver/clusterresolver.go

+		newEP := e
+		newEP.Addresses = make([]resolver.Address, len(e.Addresses))


Do we really need this newEndpoints and newEP local variables. Can we make the modifications to the individual endpoints in the endpoints slice, because that is what was happening earlier.

In the previous implementation, xdsresource.Endpoint stored raw address strings, so we were forced to construct a brand new resolver.Endpoint instance in configbuilder.priorityLocalitiesToClusterImpl().

With this refactor, the endpoints slice contains resolver.Endpoint objects that come directly from the xdsClient's cache. If we modify endpoints or its Addresses slice directly in clusterresolver, we are modifying the shared underlying memory which causes a race.

I confirmed this behavior during testing—without the explicit copy, TestConcurrentChannels fails with a data race. The newEndpoints/newEP variables ensure we are modifying a private copy.

But I have again changed the implementation. Instead of creating newEndpoints/newEP variables in clusterresolver, I'm creating a new resovler.Endpoint in configbuilder.priorityLocalitiesToClusterImpl().

add resolver.Endpoint in Endpoint struct

7720fe3

Pranjali-2501 added this to the 1.78 Release milestone Dec 6, 2025

Pranjali-2501 added Type: Feature New features or improvements in behavior Area: xDS Includes everything xDS related, including LB policies used with xDS. labels Dec 6, 2025

Pranjali-2501 self-assigned this Dec 6, 2025

fix race

ee8e198

Pranjali-2501 requested a review from easwars December 7, 2025 15:25

Pranjali-2501 assigned easwars and unassigned Pranjali-2501 Dec 7, 2025

Pranjali-2501 changed the title ~~xds: refactor xdsresource.Endpoint to embed resolver.Endpoint~~ xds: refactor xdsresource.Endpoint to embed resolver.Endpoint (gRFC A81) Dec 8, 2025

Pranjali-2501 requested a review from mbissa December 8, 2025 10:09

Pranjali-2501 assigned mbissa Dec 8, 2025

easwars reviewed Dec 8, 2025

View reviewed changes

easwars assigned Pranjali-2501 and unassigned easwars and mbissa Dec 8, 2025

Pranjali-2501 mentioned this pull request Dec 9, 2025

Get rid of xdsresource.Endpoint struct. #8757

Open

resolving comments

578caf3

Pranjali-2501 requested a review from easwars December 9, 2025 12:55

Pranjali-2501 assigned easwars and unassigned Pranjali-2501 Dec 9, 2025

		newEP := e
		newEP.Addresses = make([]resolver.Address, len(e.Addresses))

xds: refactor xdsresource.Endpoint to embed resolver.Endpoint (gRFC A81) #8750

Are you sure you want to change the base?

xds: refactor xdsresource.Endpoint to embed resolver.Endpoint (gRFC A81) #8750

Uh oh!

Conversation

Pranjali-2501 commented Dec 6, 2025 • edited by easwars Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Key Changes:

Uh oh!

codecov bot commented Dec 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Pranjali-2501 commented Dec 6, 2025 •

edited by easwars

Loading

codecov bot commented Dec 6, 2025 •

edited

Loading