fix: deep copy status in translator layer to avoid race#8778
Merged
Conversation
✅ Deploy Preview for cerulean-figolla-1f9435 canceled.
|
rudrakhp
force-pushed
the
translator_coalesce_race
branch
from
cc8f6c8 to
99a5c61
Compare
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: cc8f6c8c98
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
|
|
||
| for _, policy := range envoyPatchPolicies { | ||
| for i, orig := range envoyPatchPolicies { | ||
| envoyPatchPolicies[i] = deepCopyEnvoyPatchPolicyStatus(orig) |
There was a problem hiding this comment.
Writing the deep-copied pointer back into envoyPatchPolicies[i] still mutates the snapshot object passed from GatewayAPIResources. If a new provider update arrives while translation is running, watchable’s equality/coalescing path can read that same slice concurrently, so this assignment can still trigger the control-plane race the patch is trying to eliminate. Keep the copy in a local variable and do not modify the input slice.
Useful? React with 👍 / 👎.
| policy := &policy | ||
| // Clone the Object map to avoid racing with the watchable coalesce goroutine. | ||
| policy.Object = maps.Clone(policy.Object) | ||
| policies[policyIndex] = policy |
There was a problem hiding this comment.
policies[policyIndex] = policy mutates the shared resources.ExtensionServerPolicies snapshot before translation completes. During concurrent updates, watchable can deep-compare that same slice in another goroutine, so this write can race even after cloning policy.Object. Build status/IR from a local copied policy and leave the incoming slice unchanged.
Useful? React with 👍 / 👎.
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #8778 +/- ##
==========================================
+ Coverage 74.35% 74.39% +0.04%
==========================================
Files 245 245
Lines 38847 38951 +104
==========================================
+ Hits 28883 28978 +95
- Misses 7963 7969 +6
- Partials 2001 2004 +3 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
zirain
reviewed
Apr 17, 2026
| // Status is mutated during translation and shares a pointer with the watchable coalesce goroutine. | ||
| func deepCopyBackendStatus(in *egv1a1.Backend) *egv1a1.Backend { | ||
| out := *in | ||
| in.Status.DeepCopyInto(&out.Status) |
Member
There was a problem hiding this comment.
why not use backend.DeepCopy directly? which is better for understanding.
Member
Author
There was a problem hiding this comment.
It is less efficient than this as we only need deep copy for status which is modified. Rest of the policy can be a shallow copy. We were already doing complete deep copy before #6940 tried to optimize it.
rudrakhp
force-pushed
the
translator_coalesce_race
branch
2 times, most recently
from
c64694a to
dc37ca3
Compare
rudrakhp
force-pushed
the
translator_coalesce_race
branch
from
b0b74a1 to
dbc7ecb
Compare
Contributor
|
/retest |
arkodg
reviewed
Apr 19, 2026
| policy, found := handledPolicies[policyName] | ||
| if !found { | ||
| policy = currPolicy | ||
| policy = deepCopySecurityPolicyStatus(currPolicy) |
Contributor
There was a problem hiding this comment.
cant we just do it once in L142 ?
Member
Author
There was a problem hiding this comment.
makes sense, creating copies once in XCopiesWithStatusDeepCopy() before reusing them.
rudrakhp
force-pushed
the
translator_coalesce_race
branch
from
52c3850 to
923e3b6
Compare
Signed-off-by: Rudrakh Panigrahi <[email protected]>
rudrakhp
force-pushed
the
translator_coalesce_race
branch
from
923e3b6 to
e8199ae
Compare
zirain
approved these changes
Apr 19, 2026
rudrakhp
added a commit
that referenced
this pull request
Apr 24, 2026
* fix json report (#8614) Signed-off-by: Huabing (Robin) Zhao <[email protected]> Signed-off-by: Rudrakh Panigrahi <[email protected]> * fix: deep copy status in translator layer to avoid race (#8778) Signed-off-by: Rudrakh Panigrahi <[email protected]> * chore: setup translator test (#7627) * chore: setup gatewayapi translator test Signed-off-by: kkk777-7 <[email protected]> * remove: duplicate resources in testfiles Signed-off-by: kkk777-7 <[email protected]> * update gatewayapi test output Signed-off-by: kkk777-7 <[email protected]> --------- Signed-off-by: kkk777-7 <[email protected]> Signed-off-by: Rudrakh Panigrahi <[email protected]> * perf: introduce translator context (#7535) * perf: introduce translator context Signed-off-by: kkk777-7 <[email protected]> * perf: add policy map in translator context Signed-off-by: kkk777-7 <[email protected]> * chore: update tranlate bench Signed-off-by: kkk777-7 <[email protected]> * Revert "perf: add policy map in translator context" This reverts commit 14a3784. Signed-off-by: kkk777-7 <[email protected]> * add translator context method, remove unused function Signed-off-by: kkk777-7 <[email protected]> * update gatewayapi test output Signed-off-by: kkk777-7 <[email protected]> * add resources in translator context Signed-off-by: kkk777-7 <[email protected]> * add gatewayapi translate bench Signed-off-by: kkk777-7 <[email protected]> * fix: set resources in translator context Signed-off-by: kkk777-7 <[email protected]> * fix: go lint Signed-off-by: kkk777-7 <[email protected]> * address comments Signed-off-by: kkk777-7 <[email protected]> * feat: support egctl translate options Signed-off-by: kkk777-7 <[email protected]> * update: bench test Signed-off-by: kkk777-7 <[email protected]> * update embedded field Signed-off-by: kkk777-7 <[email protected]> * fix gatewayapi testdata output Signed-off-by: kkk777-7 <[email protected]> --------- Signed-off-by: kkk777-7 <[email protected]> Co-authored-by: zirain <[email protected]> Signed-off-by: Rudrakh Panigrahi <[email protected]> * fix gen-check Signed-off-by: Rudrakh Panigrahi <[email protected]> --------- Signed-off-by: Huabing (Robin) Zhao <[email protected]> Signed-off-by: Rudrakh Panigrahi <[email protected]> Signed-off-by: kkk777-7 <[email protected]> Co-authored-by: Huabing (Robin) Zhao <[email protected]> Co-authored-by: Kota Kimura <[email protected]> Co-authored-by: zirain <[email protected]>
skos-ninja
pushed a commit
to skos-ninja/envoy-gateway
that referenced
this pull request
May 1, 2026
) Signed-off-by: Rudrakh Panigrahi <[email protected]> Signed-off-by: Jake Oliver <[email protected]>
cnvergence
added a commit
that referenced
this pull request
May 9, 2026
* fix json report (#8614) Signed-off-by: Huabing (Robin) Zhao <[email protected]> (cherry picked from commit 4768ca7) * fix: deep copy status in translator layer to avoid race (#8778) Signed-off-by: Rudrakh Panigrahi <[email protected]> (cherry picked from commit 3f70a89) * fix: force HTTP1 for upstream connections for WS and WSS backends (#8699) * force HTTP1 for upstream connections for WS and WSS backends Signed-off-by: Huabing (Robin) Zhao <[email protected]> * use different clusters for mixed upstream protocols Signed-off-by: Huabing (Robin) Zhao <[email protected]> * update Signed-off-by: Huabing (Robin) Zhao <[email protected]> * fix lint Signed-off-by: Huabing (Robin) Zhao <[email protected]> --------- Signed-off-by: Huabing (Robin) Zhao <[email protected]> (cherry picked from commit 7633125) * fix: reason with multiple errors rejected validation (#8859) * fix: reason with multiple errors rejected validation Signed-off-by: zirain <[email protected]> * release notes Signed-off-by: zirain <[email protected]> * fix lint Signed-off-by: zirain <[email protected]> --------- Signed-off-by: zirain <[email protected]> (cherry picked from commit 7811d86) * feat(chart): Allow configuring envoy proxy image via helm chart (#8785) * feat: Allow configuring envoy proxy defaults via helm chart This commit is a continuation of the previous work to support supplying default proxy settings added in #7698 and adds three new chart values under `global.images.envoyProxy`: | Value | Type | Default | Description | |----------------------------------------|--------|------|---------------------------------------------------------------------| | `global.images.envoyProxy.image` | string | `""` | Full image name (`registry/repo:tag`) for the Envoy Proxy container | | `global.images.envoyProxy.pullPolicy` | string | `""` | Image pull policy | | `global.images.envoyProxy.pullSecrets` | list | `[]` | Image pull secrets | When any of these are set, the chart generates an `envoyProxy:` block inside the `EnvoyGateway` ConfigMap, wiring into the existing `EnvoyGatewaySpec.envoyProxy` field (added in #7698). The global `imageRegistry` override takes highest precedence, consistent with other chart components. Full EnvoyProxy defaults (replicas, resources, etc.) can be provided via `config.envoyGateway.envoyProxy`; the image values are merged on top. Closes #4764. Signed-off-by: Michael Sommerville <[email protected]> (cherry picked from commit 8570285) * add rn and version bump Signed-off-by: Karol Szwaj <[email protected]> * fix: respect backend endpoint hostname for health checks (#8929) * fix: respect backend endpoint hostname for health checks - Keep BackendTrafficPolicy HTTP health check hostnames as explicit cluster-level hosts, and leave route-derived host fallback to xDS cluster translation. - Preserve Backend endpoint hostnames as per-endpoint overrides via Endpoint.HealthCheckConfig.hostname, ahead of the route fallback. - Update gatewayapi/xDS fixtures, release notes, and generated API docs/CRDs for the host selection order. - go test ./internal/ir - go test ./internal/xds/translator - go test ./internal/gatewayapi -run TestTranslate/backendtrafficpolicy - go test ./internal/gatewayapi -run TestTranslate/(clienttrafficpolicy-http-health-check|envoyextensionpolicy-with-extproc-with-retries|envoyextensionpolicy-with-extproc-with-traffic-features|envoyproxy-accesslog-with-traffic|envoyproxy-tracing-backend-uds|envoyproxy-tracing-backend|securitypolicy-with-jwt-backendcluster|securitypolicy-with-jwt-backendsettings) - make generate - make manifests - git diff --check Signed-off-by: Arko Dasgupta <[email protected]> Co-authored-by: Codex <[email protected]> * fix gen Signed-off-by: zirain <[email protected]> --------- Signed-off-by: Arko Dasgupta <[email protected]> Signed-off-by: zirain <[email protected]> Co-authored-by: Codex <[email protected]> Co-authored-by: zirain <[email protected]> * update release notes Signed-off-by: Karol Szwaj <[email protected]> * fix gen-check Signed-off-by: Karol Szwaj <[email protected]> * Revert "feat(chart): Allow configuring envoy proxy image via helm chart (#8785)" This reverts commit 092cc67. Signed-off-by: Karol Szwaj <[email protected]> --------- Signed-off-by: Huabing (Robin) Zhao <[email protected]> Signed-off-by: Rudrakh Panigrahi <[email protected]> Signed-off-by: zirain <[email protected]> Signed-off-by: Michael Sommerville <[email protected]> Signed-off-by: Karol Szwaj <[email protected]> Signed-off-by: Arko Dasgupta <[email protected]> Co-authored-by: Huabing (Robin) Zhao <[email protected]> Co-authored-by: Rudrakh Panigrahi <[email protected]> Co-authored-by: zirain <[email protected]> Co-authored-by: Michael Sommerville <[email protected]> Co-authored-by: Arko Dasgupta <[email protected]> Co-authored-by: Codex <[email protected]>
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
What type of PR is this?
fix: deep copy status in translator layer to avoid race
What this PR does / why we need it:
DeepCopy()calls were removed in #6940 but are required to avoid race betweendeepEqual()in watchable coalesce routine and status updates in gateway api translator routine. Replaced plain shallow copy with deep copy only status.Also includes a minor spelling fix
SeverOptions->ServerOptionsWhich issue(s) this PR fixes:
Fixes #8771
Release Notes: Yes