[Bug]: OTel API version mismatch lets NoSuchMethodError escape PluginRunner
维护者通常 1 天内回复
已经有一个关联 PR 被合并了。
- #780 来自 @zhongkechen —— 已合并
评估
- 难度
- 4/5
- 预计耗时
- 3-5 天
- 新手友好度
- 35/100
- Issue 类型
- 缺陷
- 描述清晰度
- 基本清楚
- 活跃度
- 停滞
- 技术栈
- aws, java
- 领域
- backend, observability, testing
调研方向
Start with otel-plugin/src/main/java/software/amazon/lambda/durable/otel/OtelPluginSupport.java and sdk/src/main/java/software/amazon/lambda/durable/plugin/PluginRunner.java, then run the supplied B1Probe with OTel API/context 1.66.0 and 1.49.0. Done means the older-API case is safely diagnosed or contained, compatible tracing still works, regression and control coverage exist, and plugin failure does not block healthy plugins or the handler.
由索引模型根据 Issue 内容生成。
描述
Expected Behavior
An incompatible OpenTelemetry agent/API combination should produce a clear compatibility diagnostic or disable the affected instrumentation without a nonfatal linkage error escaping the plugin boundary and failing the customer invocation. Compatible configurations should continue to trace normally; fatal JVM failures need their existing separate handling.
Actual Behavior
The global-provider path calls GlobalOpenTelemetry.isSet(). If an older OpenTelemetry API is the class visible to the plugin, that method is unavailable and throws NoSuchMethodError. The published PluginRunner catches Exception, so this Error escapes.
Using the real plugin and runner, OTel API 1.66.0 returned normally; API 1.49.0 placed first on the classpath produced java.lang.NoSuchMethodError: boolean io.opentelemetry.api.GlobalOpenTelemetry.isSet() outside PluginRunner.onInvocationStart. The older API models agent/classloader version skew; this probe did not deploy an actual Java agent.
Steps to Reproduce
- Resolve
software.amazon.lambda.durable:aws-durable-execution-sdk-java-plugin-otel:2.2.1and its dependencies. - Compile
B1Probebelow in the indicated package. The package placement permits the test to mark the agent auto-configuration path as installed; this is a reproduction fixture, not recommended application code. - Run with the normal dependencies (OTel API/context 1.66.0 in the tested resolution). Observe
result=returned. - Resolve
io.opentelemetry:opentelemetry-api:1.49.0andio.opentelemetry:opentelemetry-context:1.49.0, place those two jars before the normal dependency classpath, and run the same compiled class. Observeresult=escaped_plugin_runnerand NoSuchMethodError. No AWS calls are made.
Reproduction code
package software.amazon.lambda.durable.otel;
import java.time.Instant;
import java.util.List;
import io.opentelemetry.api.GlobalOpenTelemetry;
import software.amazon.lambda.durable.plugin.InvocationInfo;
import software.amazon.lambda.durable.plugin.PluginRunner;
public class B1Probe {
public static void main(String[] args) {
OtelPluginAutoConfigurationState.markInstalled();
var plugin = new ExecutionOtelPlugin(OtelPluginConfig.builder().enableMdc(false).build());
var runner = new PluginRunner(List.of(plugin));
boolean hasMethod = java.util.Arrays.stream(GlobalOpenTelemetry.class.getMethods()).anyMatch(m -> m.getName().equals("isSet"));
try {
runner.onInvocationStart(new InvocationInfo("req", "arn:aws:lambda:us-west-2:123456789012:function:probe:1/durable-execution/probe/id",true,Instant.now()));
System.out.println("{\"apiHasIsSet\":"+hasMethod+",\"result\":\"returned\"}");
} catch (Throwable e) {
System.out.println("{\"apiHasIsSet\":"+hasMethod+",\"result\":\"escaped_plugin_runner\",\"errorType\":\""+e.getClass().getName()+"\",\"error\":\""+e.getMessage().replace("\"","\\\"")+"\"}");
}
}
}
SDK Version
Core SDK / OTel plugin 2.2.1; OTel API/context 1.66.0 control and 1.49.0 mismatch. Main rechecked at 78574e3051aac9b5ca00a0c1fd2b91eab72a9e15.
Java Version
Corretto 25.0.4.1
Is this a regression?
Unknown; no last-known-good version has been established.
Additional Context
Sources: global provider compatibility path and Exception-only plugin dispatch catch. This does not assert that every older agent version must be supported; it requests safe handling of an incompatible visible API.
The open PR #721 includes plugin lifecycle/error-isolation work and is relevant to the containment part. This issue tracks the concrete API-skew regression and supported-configuration behavior.
Acceptance criteria
- Guard/remove calls to APIs that may be absent, or safely detect and explain the required API/agent version before calling them.
- Contain nonfatal linkage failures from plugin hooks/factories without blanket-catching fatal JVM errors.
- Add the older-API regression and compatible-version control, then test the supported real agent combinations.
- Document version requirements and ensure a failed plugin does not prevent healthy plugins or the handler from completing.
- 主要语言
- Java
- 星标
- 28
- 派生
- 14
- 平均合并
- 2 天 19 小时
- 30 天内合并 PR
- 43
环境准备
- 没有 Dockerfile 或 Docker Compose 文件
- 有 Pull Request 模板
- 阅读贡献指南
从这里开始
- 先读完整个 Issue,再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库,在一个分支上完成修改。
- 提交 Pull Request,并在描述里引用这个 Issue 编号。
aws/aws-durable-execution-sdk-java 的其他 Issue
-
documentation pkg:sdk
难度 1/5 1-3 小时 新手友好度 88/100
aws/aws-durable-execution-sdk-java#645 · 1 条评论 ·
维护者通常 1 天内回复
-
enhancement
难度 2/5 1-3 小时 新手友好度 68/100
aws/aws-durable-execution-sdk-java#300 ·
维护者通常 1 天内回复
-
enhancement needs-triage
难度 5/5 一周以上 新手友好度 35/100
aws/aws-durable-execution-sdk-java#779 ·
维护者通常 1 天内回复
-
needs-triage
难度 5/5 一周以上 新手友好度 40/100
aws/aws-durable-execution-sdk-java#770 ·
维护者通常 1 天内回复
-
[Feature]: Propagate per-operation trace context for chained invokes可能已有人在做 关联的 PR 仍在进行中或已合并。 未关闭enhancement needs-triage
难度 5/5 一周以上 新手友好度 38/100
aws/aws-durable-execution-sdk-java#764 ·
维护者通常 1 天内回复
查看 aws/aws-durable-execution-sdk-java 的全部 Issue
相似的 Issue
-
[BUG] S3 CORS responses omit Access-Control-Allow-Credentials for matched origins可能已有人在做 关联的 PR 仍在进行中或已合并。 未关闭
难度 2/5 1-3 小时 新手友好度 72/100
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 72/100
sqlcipher/sqlcipher-android#97 · 1 条评论 ·
-
area-integrations
难度 2/5 1-3 小时 新手友好度 68/100
维护者通常 1 天内回复
-
bug IIIF interoperability
难度 2/5 1-3 小时 新手友好度 72/100
-
难度 2/5 1-3 小时 新手友好度 72/100