inference with timeouts leads to Internal error

[Roman Doronin]

ServingRuntime: torchserve

Current behavior

• sent requests with client timeouts (load our modelmesh)
• after some time, client starts to receive

ERROR:
  Code: Internal
  Message: org.pytorch.serve.grpc.inference.InferenceAPIsService/Predictions: INTERNAL: Model "test-search-vectorizer__isvc-409d074e44" has no worker to serve inference request. Please use scale workers API to add workers. 

• if we stop our load and start to send only one request without timeouts, we have same error or sticking

In my opinion, some resources can't released under client timeout, that lead to workers exousted. Becouse if we turn off client timeouts, there are no errors.

Expected behaivor

• requests with timeouts can't lead to workers exousted and sticking on single request to modelmesh

---

Log Info

• message

org.pytorch.serve.grpc.inference.InferenceAPIsService/Predictions: INTERNAL: Model "test-search-vectorizer__isvc-409d074e44" has no worker to serve inference request. Please use scale workers API to add workers.
InternalServerException.()

• trace

ApplierException(message:org.pytorch.serve.grpc.inference.InferenceAPIsService/Predictions: INTERNAL: Model "test-search-vectorizer__isvc-409d074e44" has no worker to serve inference request. Please use scale workers API to add workers.
InternalServerException.(), causeStacktrace:io.grpc.StatusException: INTERNAL: Model "test-search-vectorizer__isvc-409d074e44" has no worker to serve inference request. Please use scale workers API to add workers.
InternalServerException.()
	at com.ibm.watson.modelmesh.SidecarModelMesh$ExternalModel$1.onClose(SidecarModelMesh.java:450)
, grpcStatusCode:INTERNAL)
	at com.ibm.watson.modelmesh.thrift.ModelMeshService$applyModelMulti_result$applyModelMulti_resultStandardScheme.read(ModelMeshService.java:1940)
	at com.ibm.watson.modelmesh.thrift.ModelMeshService$applyModelMulti_result$applyModelMulti_resultStandardScheme.read(ModelMeshService.java:1905)
	at com.ibm.watson.modelmesh.thrift.ModelMeshService$applyModelMulti_result.read(ModelMeshService.java:1815)
	at org.apache.thrift.TServiceClient.receiveBase(TServiceClient.java:93)
	at com.ibm.watson.modelmesh.thrift.ModelMeshService$Client.recv_applyModelMulti(ModelMeshService.java:74)
	at com.ibm.watson.modelmesh.thrift.ModelMeshService$Client.applyModelMulti(ModelMeshService.java:59)
	at jdk.internal.reflect.GeneratedMethodAccessor8.invoke(Unknown Source)
	at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.base/java.lang.reflect.Method.invoke(Method.java:568)
	at com.ibm.watson.litelinks.client.ClientInvocationHandler.invoke(ClientInvocationHandler.java:496)
	at com.ibm.watson.litelinks.client.ClientInvocationHandler.invokeWithRetries(ClientInvocationHandler.java:383)
	at com.ibm.watson.litelinks.client.ClientInvocationHandler.doInvoke(ClientInvocationHandler.java:184)
	at com.ibm.watson.litelinks.client.ClientInvocationHandler.invoke(ClientInvocationHandler.java:118)
	at jdk.proxy2/jdk.proxy2.$Proxy30.applyModelMulti(Unknown Source)
	at jdk.internal.reflect.GeneratedMethodAccessor8.invoke(Unknown Source)
	at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.base/java.lang.reflect.Method.invoke(Method.java:568)
	at com.ibm.watson.modelmesh.SidecarModelMesh.invokeRemoteModel(SidecarModelMesh.java:1071)
	at com.ibm.watson.modelmesh.ModelMesh.invokeRemote(ModelMesh.java:4399)
	at com.ibm.watson.modelmesh.ModelMesh.invokeModel(ModelMesh.java:3644)
	at com.ibm.watson.modelmesh.SidecarModelMesh.callModel(SidecarModelMesh.java:1106)
	at com.ibm.watson.modelmesh.ModelMeshApi.callModel(ModelMeshApi.java:457)
	at com.ibm.watson.modelmesh.ModelMeshApi$4.onHalfClose(ModelMeshApi.java:733)
	at io.grpc.internal.ServerCallImpl$ServerStreamListenerImpl.halfClosed(ServerCallImpl.java:355)
	at io.grpc.internal.ServerImpl$JumpToApplicationThreadServerStreamListener$1HalfClosed.runInContext(ServerImpl.java:867)
	at io.grpc.internal.ContextRunnable.run(ContextRunnable.java:37)
	at io.grpc.internal.SerializingExecutor.run(SerializingExecutor.java:133)
	at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1136)
	at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:635)
	at io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)
	at java.base/java.lang.Thread.run(Thread.java:833)
	at com.ibm.watson.litelinks.server.ServerRequestThread.run(ServerRequestThread.java:47)

[Vitaly Kulpin]

Are there any updates on this bug?

[Christian Kadner]

@fsatka -- it's been a long time since you opened this issue and several updates have been made since. Can you still reproduce this bug in the latest version of ModelMesh-Serving v0.11.2?