temporalio/sdk-java

Add option to configure the deadline of `ChannelManager#getServerCapabilities`

fextr opened this issue · 6 comments

fextr commented

Expected Behavior

When I setRpcTimeout in WorkflowServiceStubsOptions it overrides the deadline in getServerCapabilitiesWithRetryOrThrow

Actual Behavior

I'm enforced to use deprecated setHealthCheckAttemptTimeout because it is still used in the ChannelManager

Steps to Reproduce the Problem

  1. Set rpcTimeout in WorkflowServiceStubsOptions.
  2. Observe that it does not override the deadline in getServerCapabilitiesWithRetryOrThrow
  3. Note the usage of the deprecated healthCheckAttemptTimeout in ChannelManager.

Specifications

  • Version: Java SDK 1.24.0
fextr commented

If that is acceptable, I want to contribute to the project and fix this issue.

Quinn-With-Two-Ns commented

getServerCapabilities is intentionally overriding the rpc timeout. If you need to configure a different deadline on getServerCapabilities it would need to be a separate option.

fextr commented

getServerCapabilities is intentionally overriding the rpc timeout. If you need to configure a different deadline on getServerCapabilities it would need to be a separate option.

Hey @Quinn-With-Two-Ns

Is your suggestion to introduce a new option, such as serverCapabilitiesTimeout, in the ServiceStubsOptions correct?

Quinn-With-Two-Ns commented

Yes, it should be called getSystemInfoTimeout though serverCapabilities is just an internal name to the Java SDK.

Quinn-With-Two-Ns commented

Why do you need to configure the deadline on getSystemInfo?

fextr commented

Why do you need to configure the deadline on getSystemInfo?

We have encountered a problem connecting to Temporal after upgrading Java SDK from version 1.23.2 to 1.24.0. When we increased the timeout problem was solved. Stack trace:

| 2024-07-01 04:59:31.284 | at io.temporal.internal.retryer.GrpcRetryer.retryWithResult(GrpcRetryer.java:60) |
| 2024-07-01 04:59:31.280 | at io.temporal.api.workflowservice.v1.WorkflowServiceGrpc$WorkflowServiceBlockingStub.getSystemInfo(WorkflowServiceGrpc.java:4645) |
| 2024-07-01 04:59:31.280 | at io.temporal.serviceclient.SystemInfoInterceptor.getServerCapabilitiesOrThrow(SystemInfoInterceptor.java:132) |
| 2024-07-01 04:59:31.280 | at io.temporal.serviceclient.SystemInfoInterceptor.lambda$getServerCapabilitiesWithRetryOrThrow$0(SystemInfoInterceptor.java:118) |
| 2024-07-01 04:59:31.280 | at io.temporal.internal.retryer.GrpcSyncRetryer.retry(GrpcSyncRetryer.java:69) |
| 2024-07-01 04:59:31.279 | io.grpc.StatusRuntimeException: DEADLINE_EXCEEDED: deadline exceeded after 0.398648053s. Name resolution delay 2.493018773 seconds. [closed=[], open=[[buffered_nanos=695520709, remote_addr=namespace-dev.etmgc.tmprl.cloud/18.157.55.99:7233]]] |
| 2024-07-01 04:59:31.279 | at io.grpc.stub.ClientCalls.toStatusRuntimeException(ClientCalls.java:268) |
| 2024-07-01 04:59:31.279 | at io.grpc.stub.ClientCalls.getUnchecked(ClientCalls.java:249) |
| 2024-07-01 04:59:31.279 | at io.grpc.stub.ClientCalls.blockingUnaryCall(ClientCalls.java:167) |
| 2024-07-01 04:59:24.584 | 04:59:24.584 [:] [main] INFO  i.t.s.WorkflowServiceStubsImpl - Created WorkflowServiceStubs for channel: ManagedChannelOrphanWrapper{delegate=ManagedChannelImpl{logId=1, target=namespace-dev.etmgc.tmprl.cloud:7233}} |

PS: Actual namespace name is different