test(network): re-add Test_conn_startSending with deterministic ordering - #4433
Closed
stevenvegt wants to merge 16 commits into
Closed
test(network): re-add Test_conn_startSending with deterministic ordering#4433stevenvegt wants to merge 16 commits into
stevenvegt wants to merge 16 commits into
Conversation
* Bump github.com/nats-io/nats-server/v2 from 2.11.12 to 2.11.15 (#4107) Bumps [github.com/nats-io/nats-server/v2](https://github.com/nats-io/nats-server) from 2.11.12 to 2.11.15. - [Release notes](https://github.com/nats-io/nats-server/releases) - [Changelog](https://github.com/nats-io/nats-server/blob/main/RELEASES.md) - [Commits](nats-io/nats-server@v2.11.12...v2.11.15) --- updated-dependencies: - dependency-name: github.com/nats-io/nats-server/v2 dependency-version: 2.11.15 dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * Release notes for v6.2.1 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Fixes GO-2026-4865 (html/template XSS), GO-2026-4866 (crypto/x509 auth bypass), GO-2026-4869 (archive/tar DoS), GO-2026-4870 (crypto/tls DoS), GO-2026-4946 (crypto/x509 DoS) Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
The v1 introspection endpoint accepted any JWT signed by a key present on the node, allowing VP JWTs to be replayed as access tokens. This adds three validation checks to IntrospectAccessToken: - Require typ header to be "at+jwt" (set on issuance, verified on introspection) - Require non-empty iss, sub, and service claims - Verify iss claim matches the DID extracted from the kid header Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Add v6.2.3 release notes --------- Co-authored-by: Steven van der Vegt <steven.vandervegt@s-curve.nl> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Probably due to a new version of vault, the e2e test failed. Vault tried to set setcap in vault container for CI compatibility which is not allowed by the CI runner. Created the correct container dependencies and skip the setcap for vault fixed the problem. Co-authored-by: Steven van der Vegt <steven.vandervegt@s-curve.nl>
Updates go-stoabs from v1.11.0 to v1.11.1, which removes an unnecessary Go-level read lock from the BBolt wrapper. The read lock caused reader starvation (context deadline exceeded) when a write transaction was pending, preventing network notifiers from starting on nodes with slow disk I/O (e.g. SMB/Azure Files volume mounts). Backport of #4170 to V6.2. Fixes #4162 Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* Bump alpine from 3.23.3 to 3.23.4 (#4203) Bumps alpine from 3.23.3 to 3.23.4. --- updated-dependencies: - dependency-name: alpine dependency-version: 3.23.4 dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * Add v6.2.5 release notes --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
…2.5) (#4239) * Upgrade Go to 1.26.3 and golang.org/x/net to v0.53.0 to fix CVEs (v6.2.6) Assisted by AI * Fold CVE fixes into pending v6.2.5 release v6.2.5 was prepared (alpine bump) but never tagged. Combine with the Go and golang.org/x/net upgrades and ship under v6.2.5 rather than introducing a v6.2.6 gap. Assisted by AI
* Exclude retraction presentations from Discovery Service search (#4193) Retraction markers are stored on the timeline so Get() can replicate them, but must not surface from Search(). With an empty query, no credential join was applied, so retraction VPs (which have no credentials) were returned to API callers. Fixes #4192 Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Make HTTPErrorHandler idempotent on committed responses (#4243) Echo's BodyDump middleware (used when http.log: metadata-and-body) calls c.Error(err) on its way out and still returns the err, so echo's server loop invokes HTTPErrorHandler a second time for the same request. The first invocation wrote the response correctly; the second logged the operation error a second time and warned "Unable to send error back to client, response already committed" — visible noise without a real problem. echo.DefaultHTTPErrorHandler short-circuits on Committed=true; ours did not. Mirror that behavior. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * Add v6.2.6 release notes Assisted by AI * Update release_notes.rst --------- Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Co-authored-by: Steven van der Vegt <steven.vandervegt@s-curve.nl>
…ters to fix CVEs (v6.2.7) (#4282) * Upgrade golang.org/x/crypto and golang.org/x/net to fix CVEs (v6.2.7) - golang.org/x/crypto v0.50.0 -> v0.52.0 (GO-2026-5018) - golang.org/x/net v0.53.0 -> v0.55.0 (GO-2026-5026) Assisted by AI * Upgrade go-jose and OTLP HTTP exporters to fix CVEs (v6.2.7) - github.com/go-jose/go-jose/v4 v4.1.3 -> v4.1.4 (GO-2026-4945) - go.opentelemetry.io/otel/exporters/otlp/otlplog/otlploghttp v0.16.0 -> v0.19.0 (GO-2026-4985) - go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.42.0 -> v1.43.0 (GO-2026-4985) Assisted by AI
Assisted by AI
Backport of #4250 and #4259 from master, limited to CI: - add .github/workflows/go-test.yaml, remove .circleci/config.yml - build-images: build only linux/amd64 on PRs, multi-arch on push/tag - e2e-tests: enable GHA layer cache - drop brittle DNS-error / readdir-order test assertions exposed by the ubuntu-latest runner (same fixes as #4250) so go-test passes Assisted by AI
…g down (#4276) (#4381) Test_grpcConnectionManager_Peers/inbound_stream_triggers_observer and the mirror outbound subtest set Authenticate expectations on both peer's authenticator mocks but only waited for one side's observer to fire before calling cm2.Stop(). When the observed side won the race, the other side's authenticator could still be in flight when t.Cleanup ran gomock's Finish(), reporting a missing call. Extend the wait condition to also require the unobserved side's Peers() list to be populated, which only happens after that side's Authenticate returns. Assisted-by: AI Co-authored-by: Steven van der Vegt <steven.vandervegt@s-curve.nl>
Assisted by AI
The test was removed on master in #2618 because it was flaky: it cancelled the stream before calling disconnect(), racing the receive goroutine (which stores the stream error as close status) against disconnect() cancelling the connection context. Disconnecting first guarantees the context is cancelled before RecvMsg returns, making the test deterministic while restoring the goroutine-exit and no-panic coverage. Verified with -race -count=100. Assisted-by: AI
stevenvegt
requested review from
Dirklectisch,
JorisHeadease,
gerardsn,
reinkrul and
woutslakhorst
as code owners
July 27, 2026 16:39
Member
Author
|
Wrong head branch (created from the V6.2 fix branch); replaced by a PR from fix/flaky-conn-test-master. |
Contributor
22 new issues
|
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
2 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.
Summary
Re-adds
Test_conn_startSending/disconnect_does_not_panic, which was removed in #2618 because it was flaky. The flake was one over-specified assertion: the test cancelled the stream before callingdisconnect(), racing the receive goroutine (which stores the stream error as close status) againstdisconnect()cancelling the connection context.Disconnecting first makes the test deterministic and restores coverage for goroutine exit and the disconnect panic guard (#1017). Companion PRs: same re-add on V6.2, reorder-only fix on V5.4 (which still had the racy version and flaked on #4425).
Verified with
go test -race -count=100: 100/100 pass.Assisted-by: AI