Repository navigation
test(spring-boot): [Sample Tests 19] Move the Boot 2 WebFlux sample's tests into the module - #6311
Merged
runningcode merged 4 commits intoOct 23, 2026
Conversation
This was referenced Oct 20, 2026
Merged
📲 Install BuildsAndroid
|
This was referenced Oct 20, 2026
Merged
Draft
runningcode
force-pushed
the
no/boot4-webflux-integration-test
branch
from
October 22, 2026 16:02
d618a3e to
8583f1f
Compare
runningcode
force-pushed
the
no/boot2-webflux-integration-test
branch
from
October 23, 2026 08:24
b1b2e95 to
ff23062
Compare
runningcode
force-pushed
the
no/boot4-webflux-integration-test
branch
from
October 23, 2026 08:58
2981d31 to
a261b1f
Compare
runningcode
force-pushed
the
no/boot2-webflux-integration-test
branch
2 times, most recently
from
October 23, 2026 09:35
78704b4 to
a000438
Compare
runningcode
force-pushed
the
no/boot4-webflux-integration-test
branch
2 times, most recently
from
October 23, 2026 11:51
9cd5546 to
a5cdced
Compare
runningcode
force-pushed
the
no/boot2-webflux-integration-test
branch
3 times, most recently
from
October 23, 2026 13:02
b4103c8 to
761dd4c
Compare
runningcode
force-pushed
the
no/boot4-webflux-integration-test
branch
from
October 23, 2026 13:02
3a815e4 to
4cfc49d
Compare
…dule Adds an integrationTestWebflux suite to sentry-spring-boot that mirrors the Boot 3 one, without the Reactor thread-local accessor that Spring Boot 2's Reactor lacks. The GraphQL greeting lives in its own file so the suite can leave it out below Spring Boot 2.7, like the servlet suite's GraphQL. The Boot 2 WebFlux sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 2 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
…ore Metrics Ports the tracing checks and the plain test client from the servlet suite, and runs the WebFlux sample's Metrics tests in the reactive app too, rather than only in the servlet suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
…lver The sample's resolver returned a Mono, so its test covered Sentry finishing the field's span when the Mono completes. The port returned a String, which only covers the synchronous path. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
runningcode
force-pushed
the
no/boot4-webflux-integration-test
branch
from
October 23, 2026 13:53
4cfc49d to
953d39a
Compare
runningcode
force-pushed
the
no/boot2-webflux-integration-test
branch
from
October 23, 2026 13:53
761dd4c to
897815a
Compare
runningcode
marked this pull request as ready for review
October 23, 2026 13:57
runningcode
requested review from
0xadam-brown,
adinauer,
markushi and
romtsn
as code owners
October 23, 2026 13:57
0xadam-brown
approved these changes
Oct 23, 2026
…mple's tests (#6312) * test(spring-boot): Move the Boot 3 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot-jakarta. The app traces through the OpenTelemetry SDK in its own JVM, set up by sentry-opentelemetry-agentless-spring, and the suite ports the no-agent sample's tests: spans from the OTel annotation, the OTel API and the Sentry API with their feature flags, the OTel database span, trace continuation, client spans, cache spans, GraphQL, and Kafka traced by OTel instead of Sentry's Spring Kafka integration. The sample's Metrics tests were copies of the servlet ones and don't depend on tracing. The suite needs a TestRestTemplate built outside the context: the one Spring Boot provides is instrumented by the OTel starter, which replaced the trace headers the tests send and started a new trace. Sentry exports OTel spans in batches every 5 seconds, so every test waited that long. MockSentryServer gets a hook that runs before each check while waiting, which the suite uses to flush the tracer provider, taking the suite from about 112s to 8s of test time. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 3 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Rejoin a wrapped comment in the Boot 3 OTel no-agent suite Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 3 no-agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 21] Move the Boot 4 OTel no-agent sample's tests (#6313) * test(spring-boot): Move the Boot 4 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot-4, the Boot 4 counterpart of the Boot 3 one: the app traces through the OpenTelemetry SDK in its own JVM, set up by sentry-opentelemetry-agentless-spring. The sample's tests were the same as the Boot 3 sample's, minus Cache, so the suite is the Boot 3 one with Boot 4's packages and no Cache test. Its Metrics tests were copies of the servlet ones and don't depend on tracing. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 4 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 4 no-agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 22] Move the Boot 2 OTel no-agent sample's tests (#6314) * test(spring-boot): Move the Boot 2 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot, the Boot 2 counterpart of the Boot 3 one. The sample's tests were the Boot 3 sample's minus Cache and RestClient, which Boot 2 doesn't have. GraphQL and Kafka are gated on Boot 2.7, as in the servlet suite, and Kafka gets a context of its own. Sentry is global, so a second live context re-initializes it with that context's OpenTelemetry SDK, and the Sentry API spans of the first context's tests went to the wrong one. Each test class therefore starts its own context and closes it afterwards. The sample has no system tests left, so it comes off the system-test runner and the Spring Boot 2 matrix, which now runs the new suite, and the backend workflow's base row moves to the console sample. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Port the tracing and GraphQL checks to Boot 2 no-agent and restore Metrics Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 23] Port the Boot 3 OTel agent sample's Person, tracing and Todo tests (#6315) * test(spring-boot): Port the Boot 3 OTel agent sample's Person, tracing and Todo tests The agent tests now start the fixture app once per test JVM and share it, because it takes seconds to start. Which mode initializes Sentry is a property of the test task, so the suite's task runs with the agent initializing it and a new integrationTestOtelAgentWithoutAutoInit task runs the same tests with the Spring Boot starter doing it. Gradle runs the two in parallel. The fixture gains the Person POST, distributed tracing and Todo endpoints, and the suite ports the agent sample's Person, DistributedTracing and Todo tests. Sentry exports the agent's spans in batches every 5 seconds, which can't be configured, so each test waits for the next batch and the negative tracing check waits past one. ChildJvmApp now copies the app's output into the test's as it arrives instead of printing it on close. The shared app is closed when the JVM exits, where its whole log would otherwise land on the console. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to the Boot 3 agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 24] Port the rest of the Boot 3 OTel agent sample's tests (#6316) * test(spring-boot): Port the rest of the Boot 3 OTel agent sample's tests The agent fixture gains the Cache, Metrics, GraphQL and Kafka endpoints, and the agent suite ports the sample's remaining tests in both init modes. The agent traces each GraphQL operation as one span named after it, so the queries are named. Kafka uses an in-JVM KRaft broker that the test JVM starts next to the app. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 3 matrix. The matrix no longer builds the agent separately, because the suites build the jar they attach. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Check the Boot 3 agent GraphQL task query's assignee and creator ids Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
runningcode
merged commit Oct 23, 2026
d9b9fe9
into
no/boot4-webflux-integration-test
44 checks passed
runningcode
added a commit
that referenced
this pull request
Oct 23, 2026
… tests into the module (#6310) * test(spring-boot): Move the Boot 4 WebFlux sample's tests into the module Adds an integrationTestWebflux suite to sentry-spring-boot-4 that mirrors the Boot 3 one, adjusted for Spring Boot 4: the webclient starter for WebClient.Builder, the restclient starter that TestRestTemplate needs, Jackson 3, and @AutoConfigureTestRestTemplate. The Todo test reads the port with @LocalServerPort because TestRestTemplate's root URI doesn't hold the port on Boot 4.0. The Boot 4 WebFlux sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 4 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 4 WebFlux Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Return a Mono from the Boot 4 WebFlux GraphQL resolver The sample's resolver returned a Mono, so its test covered Sentry finishing the field's span when the Mono completes. The port returned a String, which only covers the synchronous path. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 19] Move the Boot 2 WebFlux sample's tests into the module (#6311) * test(spring-boot): Move the Boot 2 WebFlux sample's tests into the module Adds an integrationTestWebflux suite to sentry-spring-boot that mirrors the Boot 3 one, without the Reactor thread-local accessor that Spring Boot 2's Reactor lacks. The GraphQL greeting lives in its own file so the suite can leave it out below Spring Boot 2.7, like the servlet suite's GraphQL. The Boot 2 WebFlux sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 2 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Tighten the Boot 2 WebFlux tracing checks and restore Metrics Ports the tracing checks and the plain test client from the servlet suite, and runs the WebFlux sample's Metrics tests in the reactive app too, rather than only in the servlet suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Return a Mono from the Boot 2 WebFlux GraphQL resolver The sample's resolver returned a Mono, so its test covered Sentry finishing the field's span when the Mono completes. The port returned a String, which only covers the synchronous path. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 20] Move the Boot 3 OTel no-agent sample's tests (#6312) * test(spring-boot): Move the Boot 3 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot-jakarta. The app traces through the OpenTelemetry SDK in its own JVM, set up by sentry-opentelemetry-agentless-spring, and the suite ports the no-agent sample's tests: spans from the OTel annotation, the OTel API and the Sentry API with their feature flags, the OTel database span, trace continuation, client spans, cache spans, GraphQL, and Kafka traced by OTel instead of Sentry's Spring Kafka integration. The sample's Metrics tests were copies of the servlet ones and don't depend on tracing. The suite needs a TestRestTemplate built outside the context: the one Spring Boot provides is instrumented by the OTel starter, which replaced the trace headers the tests send and started a new trace. Sentry exports OTel spans in batches every 5 seconds, so every test waited that long. MockSentryServer gets a hook that runs before each check while waiting, which the suite uses to flush the tracer provider, taking the suite from about 112s to 8s of test time. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 3 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Rejoin a wrapped comment in the Boot 3 OTel no-agent suite Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 3 no-agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 21] Move the Boot 4 OTel no-agent sample's tests (#6313) * test(spring-boot): Move the Boot 4 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot-4, the Boot 4 counterpart of the Boot 3 one: the app traces through the OpenTelemetry SDK in its own JVM, set up by sentry-opentelemetry-agentless-spring. The sample's tests were the same as the Boot 3 sample's, minus Cache, so the suite is the Boot 3 one with Boot 4's packages and no Cache test. Its Metrics tests were copies of the servlet ones and don't depend on tracing. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 4 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 4 no-agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 22] Move the Boot 2 OTel no-agent sample's tests (#6314) * test(spring-boot): Move the Boot 2 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot, the Boot 2 counterpart of the Boot 3 one. The sample's tests were the Boot 3 sample's minus Cache and RestClient, which Boot 2 doesn't have. GraphQL and Kafka are gated on Boot 2.7, as in the servlet suite, and Kafka gets a context of its own. Sentry is global, so a second live context re-initializes it with that context's OpenTelemetry SDK, and the Sentry API spans of the first context's tests went to the wrong one. Each test class therefore starts its own context and closes it afterwards. The sample has no system tests left, so it comes off the system-test runner and the Spring Boot 2 matrix, which now runs the new suite, and the backend workflow's base row moves to the console sample. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Port the tracing and GraphQL checks to Boot 2 no-agent and restore Metrics Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 23] Port the Boot 3 OTel agent sample's Person, tracing and Todo tests (#6315) * test(spring-boot): Port the Boot 3 OTel agent sample's Person, tracing and Todo tests The agent tests now start the fixture app once per test JVM and share it, because it takes seconds to start. Which mode initializes Sentry is a property of the test task, so the suite's task runs with the agent initializing it and a new integrationTestOtelAgentWithoutAutoInit task runs the same tests with the Spring Boot starter doing it. Gradle runs the two in parallel. The fixture gains the Person POST, distributed tracing and Todo endpoints, and the suite ports the agent sample's Person, DistributedTracing and Todo tests. Sentry exports the agent's spans in batches every 5 seconds, which can't be configured, so each test waits for the next batch and the negative tracing check waits past one. ChildJvmApp now copies the app's output into the test's as it arrives instead of printing it on close. The shared app is closed when the JVM exits, where its whole log would otherwise land on the console. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to the Boot 3 agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 24] Port the rest of the Boot 3 OTel agent sample's tests (#6316) * test(spring-boot): Port the rest of the Boot 3 OTel agent sample's tests The agent fixture gains the Cache, Metrics, GraphQL and Kafka endpoints, and the agent suite ports the sample's remaining tests in both init modes. The agent traces each GraphQL operation as one span named after it, so the queries are named. Kafka uses an in-JVM KRaft broker that the test JVM starts next to the app. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 3 matrix. The matrix no longer builds the agent separately, because the suites build the jar they attach. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Check the Boot 3 agent GraphQL task query's assignee and creator ids Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
runningcode
added a commit
that referenced
this pull request
Oct 23, 2026
… tests into the module (#6308) * test(spring-boot): Move the Boot 3 WebFlux sample's tests into the module Adds an integrationTestWebflux suite to sentry-spring-boot-jakarta. It runs a reactive app without the servlet stack, so Spring Boot starts a WebFlux server, and ports the WebFlux sample's Person, DistributedTracing, Todo and GraphQL tests. The sample's Cache and Metrics tests were copies of the servlet ones and don't depend on the web stack, so the integrationTest suite already covers them. The WebFlux sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 3 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Tighten the Boot 3 WebFlux tracing checks and restore Cache and Metrics Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Return a Mono from the Boot 3 WebFlux GraphQL resolver The sample's resolver returned a Mono, so its test covered Sentry finishing the field's span when the Mono completes. The port returned a String, which only covers the synchronous path. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 18] Move the Boot 4 WebFlux sample's tests into the module (#6310) * test(spring-boot): Move the Boot 4 WebFlux sample's tests into the module Adds an integrationTestWebflux suite to sentry-spring-boot-4 that mirrors the Boot 3 one, adjusted for Spring Boot 4: the webclient starter for WebClient.Builder, the restclient starter that TestRestTemplate needs, Jackson 3, and @AutoConfigureTestRestTemplate. The Todo test reads the port with @LocalServerPort because TestRestTemplate's root URI doesn't hold the port on Boot 4.0. The Boot 4 WebFlux sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 4 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 4 WebFlux Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Return a Mono from the Boot 4 WebFlux GraphQL resolver The sample's resolver returned a Mono, so its test covered Sentry finishing the field's span when the Mono completes. The port returned a String, which only covers the synchronous path. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 19] Move the Boot 2 WebFlux sample's tests into the module (#6311) * test(spring-boot): Move the Boot 2 WebFlux sample's tests into the module Adds an integrationTestWebflux suite to sentry-spring-boot that mirrors the Boot 3 one, without the Reactor thread-local accessor that Spring Boot 2's Reactor lacks. The GraphQL greeting lives in its own file so the suite can leave it out below Spring Boot 2.7, like the servlet suite's GraphQL. The Boot 2 WebFlux sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 2 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Tighten the Boot 2 WebFlux tracing checks and restore Metrics Ports the tracing checks and the plain test client from the servlet suite, and runs the WebFlux sample's Metrics tests in the reactive app too, rather than only in the servlet suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Return a Mono from the Boot 2 WebFlux GraphQL resolver The sample's resolver returned a Mono, so its test covered Sentry finishing the field's span when the Mono completes. The port returned a String, which only covers the synchronous path. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 20] Move the Boot 3 OTel no-agent sample's tests (#6312) * test(spring-boot): Move the Boot 3 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot-jakarta. The app traces through the OpenTelemetry SDK in its own JVM, set up by sentry-opentelemetry-agentless-spring, and the suite ports the no-agent sample's tests: spans from the OTel annotation, the OTel API and the Sentry API with their feature flags, the OTel database span, trace continuation, client spans, cache spans, GraphQL, and Kafka traced by OTel instead of Sentry's Spring Kafka integration. The sample's Metrics tests were copies of the servlet ones and don't depend on tracing. The suite needs a TestRestTemplate built outside the context: the one Spring Boot provides is instrumented by the OTel starter, which replaced the trace headers the tests send and started a new trace. Sentry exports OTel spans in batches every 5 seconds, so every test waited that long. MockSentryServer gets a hook that runs before each check while waiting, which the suite uses to flush the tracer provider, taking the suite from about 112s to 8s of test time. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 3 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Rejoin a wrapped comment in the Boot 3 OTel no-agent suite Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 3 no-agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 21] Move the Boot 4 OTel no-agent sample's tests (#6313) * test(spring-boot): Move the Boot 4 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot-4, the Boot 4 counterpart of the Boot 3 one: the app traces through the OpenTelemetry SDK in its own JVM, set up by sentry-opentelemetry-agentless-spring. The sample's tests were the same as the Boot 3 sample's, minus Cache, so the suite is the Boot 3 one with Boot 4's packages and no Cache test. Its Metrics tests were copies of the servlet ones and don't depend on tracing. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 4 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 4 no-agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 22] Move the Boot 2 OTel no-agent sample's tests (#6314) * test(spring-boot): Move the Boot 2 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot, the Boot 2 counterpart of the Boot 3 one. The sample's tests were the Boot 3 sample's minus Cache and RestClient, which Boot 2 doesn't have. GraphQL and Kafka are gated on Boot 2.7, as in the servlet suite, and Kafka gets a context of its own. Sentry is global, so a second live context re-initializes it with that context's OpenTelemetry SDK, and the Sentry API spans of the first context's tests went to the wrong one. Each test class therefore starts its own context and closes it afterwards. The sample has no system tests left, so it comes off the system-test runner and the Spring Boot 2 matrix, which now runs the new suite, and the backend workflow's base row moves to the console sample. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Port the tracing and GraphQL checks to Boot 2 no-agent and restore Metrics Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 23] Port the Boot 3 OTel agent sample's Person, tracing and Todo tests (#6315) * test(spring-boot): Port the Boot 3 OTel agent sample's Person, tracing and Todo tests The agent tests now start the fixture app once per test JVM and share it, because it takes seconds to start. Which mode initializes Sentry is a property of the test task, so the suite's task runs with the agent initializing it and a new integrationTestOtelAgentWithoutAutoInit task runs the same tests with the Spring Boot starter doing it. Gradle runs the two in parallel. The fixture gains the Person POST, distributed tracing and Todo endpoints, and the suite ports the agent sample's Person, DistributedTracing and Todo tests. Sentry exports the agent's spans in batches every 5 seconds, which can't be configured, so each test waits for the next batch and the negative tracing check waits past one. ChildJvmApp now copies the app's output into the test's as it arrives instead of printing it on close. The shared app is closed when the JVM exits, where its whole log would otherwise land on the console. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to the Boot 3 agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 24] Port the rest of the Boot 3 OTel agent sample's tests (#6316) * test(spring-boot): Port the rest of the Boot 3 OTel agent sample's tests The agent fixture gains the Cache, Metrics, GraphQL and Kafka endpoints, and the agent suite ports the sample's remaining tests in both init modes. The agent traces each GraphQL operation as one span named after it, so the queries are named. Kafka uses an in-JVM KRaft broker that the test JVM starts next to the app. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 3 matrix. The matrix no longer builds the agent separately, because the suites build the jar they attach. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Check the Boot 3 agent GraphQL task query's assignee and creator ids Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 1 potential issue.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit 899f63a. Configure here.
| classpath = sources.runtimeClasspath | ||
| useJUnit() | ||
| runAgentFixture(autoInit = false) | ||
| } |
There was a problem hiding this comment.
Agent auto-init-false tests skip check
Medium Severity
integrationTestOtelAgentWithoutAutoInit is a standalone Test task, so Gradle never adds it to check. The old suite ran both agent init modes in one task; now SENTRY_AUTO_INIT=false only runs when that task is named, so default verification misses that path.
Triggered by project rule: PR Review Guidelines for Cursor Bugbot
Reviewed by Cursor Bugbot for commit 899f63a. Configure here.
runningcode
added a commit
that referenced
this pull request
Oct 23, 2026
… system tests with a Log4j2 suite (#6306) * test(spring-boot): Remove the log4j2 Boot samples' system tests The jakarta-log4j2 and 4-log4j2 samples' system tests were copies of the plain samples' tests, which now run in the modules' integrationTest suites. The only addition was a check that a logged error carries the Log4j integration. That's covered by SentryLog4j2AppenderAutoConfigurationTest, which checks the appender Spring Boot attaches, and by sentry-log4j2's own tests of what the appender sends. Both samples come off the system-test runner and the backend workflow. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Test the auto-configured Log4j2 appender end to end The log4j2 Boot samples' tests were removed as covered elsewhere, but nothing else sends an event through the Log4j2 appender that Spring Boot's auto-configuration attaches. integrationTestLog4j2, in sentry-spring-boot-jakarta and sentry-spring-boot-4, runs a servlet app with Log4j2 instead of Logback and checks that an error logged through SLF4J reaches Sentry with the request's feature flag and the Log4j integration, as the samples' Person test did. Gradle's module replacement swaps the starters, as Spring Boot documents. It only acts on a classpath with both, which only this suite has. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 17] Move the Boot 3 WebFlux sample's tests into the module (#6308) * test(spring-boot): Move the Boot 3 WebFlux sample's tests into the module Adds an integrationTestWebflux suite to sentry-spring-boot-jakarta. It runs a reactive app without the servlet stack, so Spring Boot starts a WebFlux server, and ports the WebFlux sample's Person, DistributedTracing, Todo and GraphQL tests. The sample's Cache and Metrics tests were copies of the servlet ones and don't depend on the web stack, so the integrationTest suite already covers them. The WebFlux sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 3 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Tighten the Boot 3 WebFlux tracing checks and restore Cache and Metrics Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Return a Mono from the Boot 3 WebFlux GraphQL resolver The sample's resolver returned a Mono, so its test covered Sentry finishing the field's span when the Mono completes. The port returned a String, which only covers the synchronous path. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 18] Move the Boot 4 WebFlux sample's tests into the module (#6310) * test(spring-boot): Move the Boot 4 WebFlux sample's tests into the module Adds an integrationTestWebflux suite to sentry-spring-boot-4 that mirrors the Boot 3 one, adjusted for Spring Boot 4: the webclient starter for WebClient.Builder, the restclient starter that TestRestTemplate needs, Jackson 3, and @AutoConfigureTestRestTemplate. The Todo test reads the port with @LocalServerPort because TestRestTemplate's root URI doesn't hold the port on Boot 4.0. The Boot 4 WebFlux sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 4 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 4 WebFlux Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Return a Mono from the Boot 4 WebFlux GraphQL resolver The sample's resolver returned a Mono, so its test covered Sentry finishing the field's span when the Mono completes. The port returned a String, which only covers the synchronous path. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 19] Move the Boot 2 WebFlux sample's tests into the module (#6311) * test(spring-boot): Move the Boot 2 WebFlux sample's tests into the module Adds an integrationTestWebflux suite to sentry-spring-boot that mirrors the Boot 3 one, without the Reactor thread-local accessor that Spring Boot 2's Reactor lacks. The GraphQL greeting lives in its own file so the suite can leave it out below Spring Boot 2.7, like the servlet suite's GraphQL. The Boot 2 WebFlux sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 2 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Tighten the Boot 2 WebFlux tracing checks and restore Metrics Ports the tracing checks and the plain test client from the servlet suite, and runs the WebFlux sample's Metrics tests in the reactive app too, rather than only in the servlet suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Return a Mono from the Boot 2 WebFlux GraphQL resolver The sample's resolver returned a Mono, so its test covered Sentry finishing the field's span when the Mono completes. The port returned a String, which only covers the synchronous path. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 20] Move the Boot 3 OTel no-agent sample's tests (#6312) * test(spring-boot): Move the Boot 3 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot-jakarta. The app traces through the OpenTelemetry SDK in its own JVM, set up by sentry-opentelemetry-agentless-spring, and the suite ports the no-agent sample's tests: spans from the OTel annotation, the OTel API and the Sentry API with their feature flags, the OTel database span, trace continuation, client spans, cache spans, GraphQL, and Kafka traced by OTel instead of Sentry's Spring Kafka integration. The sample's Metrics tests were copies of the servlet ones and don't depend on tracing. The suite needs a TestRestTemplate built outside the context: the one Spring Boot provides is instrumented by the OTel starter, which replaced the trace headers the tests send and started a new trace. Sentry exports OTel spans in batches every 5 seconds, so every test waited that long. MockSentryServer gets a hook that runs before each check while waiting, which the suite uses to flush the tracer provider, taking the suite from about 112s to 8s of test time. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 3 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Rejoin a wrapped comment in the Boot 3 OTel no-agent suite Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 3 no-agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 21] Move the Boot 4 OTel no-agent sample's tests (#6313) * test(spring-boot): Move the Boot 4 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot-4, the Boot 4 counterpart of the Boot 3 one: the app traces through the OpenTelemetry SDK in its own JVM, set up by sentry-opentelemetry-agentless-spring. The sample's tests were the same as the Boot 3 sample's, minus Cache, so the suite is the Boot 3 one with Boot 4's packages and no Cache test. Its Metrics tests were copies of the servlet ones and don't depend on tracing. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 4 matrix, which now runs the new suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to Boot 4 no-agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 22] Move the Boot 2 OTel no-agent sample's tests (#6314) * test(spring-boot): Move the Boot 2 OTel no-agent sample's tests into the module Adds an integrationTestOtelNoAgent suite to sentry-spring-boot, the Boot 2 counterpart of the Boot 3 one. The sample's tests were the Boot 3 sample's minus Cache and RestClient, which Boot 2 doesn't have. GraphQL and Kafka are gated on Boot 2.7, as in the servlet suite, and Kafka gets a context of its own. Sentry is global, so a second live context re-initializes it with that context's OpenTelemetry SDK, and the Sentry API spans of the first context's tests went to the wrong one. Each test class therefore starts its own context and closes it afterwards. The sample has no system tests left, so it comes off the system-test runner and the Spring Boot 2 matrix, which now runs the new suite, and the backend workflow's base row moves to the console sample. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Port the tracing and GraphQL checks to Boot 2 no-agent and restore Metrics Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 23] Port the Boot 3 OTel agent sample's Person, tracing and Todo tests (#6315) * test(spring-boot): Port the Boot 3 OTel agent sample's Person, tracing and Todo tests The agent tests now start the fixture app once per test JVM and share it, because it takes seconds to start. Which mode initializes Sentry is a property of the test task, so the suite's task runs with the agent initializing it and a new integrationTestOtelAgentWithoutAutoInit task runs the same tests with the Spring Boot starter doing it. Gradle runs the two in parallel. The fixture gains the Person POST, distributed tracing and Todo endpoints, and the suite ports the agent sample's Person, DistributedTracing and Todo tests. Sentry exports the agent's spans in batches every 5 seconds, which can't be configured, so each test waits for the next batch and the negative tracing check waits past one. ChildJvmApp now copies the app's output into the test's as it arrives instead of printing it on close. The shared app is closed when the JVM exits, where its whole log would otherwise land on the console. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test: Port the audit fixes to the Boot 3 agent Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): [Sample Tests 24] Port the rest of the Boot 3 OTel agent sample's tests (#6316) * test(spring-boot): Port the rest of the Boot 3 OTel agent sample's tests The agent fixture gains the Cache, Metrics, GraphQL and Kafka endpoints, and the agent suite ports the sample's remaining tests in both init modes. The agent traces each GraphQL operation as one span named after it, so the queries are named. Kafka uses an in-JVM KRaft broker that the test JVM starts next to the app. The sample has no system tests left, so it comes off the system-test runner, the backend workflow and the Spring Boot 3 matrix. The matrix no longer builds the agent separately, because the suites build the jar they attach. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * test(spring-boot): Check the Boot 3 agent GraphQL task query's assignee and creator ids Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
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
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.


PR Stack (Sample Tests)
main)📜 Description
Adds an
integrationTestWebfluxsuite tosentry-spring-boot, a copy of the Boot 3 one from #6308. It ports the Boot 2 WebFlux sample's tests (Person, DistributedTracing, Todo, and the GraphQL greeting) and runs them on a reactive Netty server.Changes for Spring Boot 2:
sentry.reactive.thread-local-accessor-enabled, matching the Boot 2 sample. Boot 2's Reactor 3.4 has no automatic context propagation.GreetingController.kt. Below 2.7, the existingbootSupportsGraphqlAndKafkaflag leaves it and the GraphQL test out, as it does for the servlet suite.The sample's Metrics copy runs in this suite too, as in #6308. With no system tests left, the sample comes off the runner, the backend workflow and the Spring Boot 2 matrix, which now runs the new suite.
This finishes the WebFlux flavor: none of the three WebFlux samples has system tests now.
After a coverage audit against the sample's tests:
sample_randand deferred-sampling tests check the envelope header's trace ID, and a test client built outside the context, as in test(spring-boot): [Sample Tests 7] Move the Boot 3 tracing and HTTP client tests #6293, and restores the Metrics tests, as in test(spring-boot): [Sample Tests 17] Move the Boot 3 WebFlux sample's tests into the module #6308.💡 Motivation and Context
This continues moving the sample system tests into their modules, here for the WebFlux flavor on Boot 2.
💚 How did you test it?
:sentry-spring-boot:checkpasses.📝 Checklist
sendDefaultPIIis enabled.🔮 Next steps
The OTel no-agent samples, then the Spring WAR samples.
#skip-changelog