Yu Xia

Init internal cluster client using membership info

1

**What changed?** Init inner cluster client using membership info **Why?** Inner cluster client doesn't need to use cluster public endpoint. **How did you test it?** Existing tests. **Potential risks** **Is...

**What changed?** Move cluster management api to operator handler. **Why?** The cluster management APIs are access by tctl which will depend on operator client later. For describe cluster API, The...

## Expected Behavior calculate namespace replication ack level by active cluster connections NS replication metadata: Cluster A (disconnected): 100 Cluster B (connected): 1000 Cluster C (connected): 500 minimum ack level:...

Update frontend drain time

1

**What changed?** Update frontend drain time from minimum drain to dynamic config duration **Why?** This allows users to config the service draining duration. **How did you test it?** **Potential risks**...

**Is your feature request related to a problem? Please describe.** To detect non-deterministic errors with the unit test framework **Proposed Solution** This is a draft idea about the unit testing....

Reset workflows contains child workflow cause non-deterministic error

3

**Describe the bug** Reset workflows contains child workflow cause non-deterministic error if the child workflow id is not set. https://github.com/uber-go/cadence-client/blob/5cc879bc3a9b2258ca569b5cb971231d9de0ff9b/internal/internal_event_handlers.go#L357 **To Reproduce** Is the issue reproducible? - Yes **Steps to...

We see NPE in ReplayDecider with line: decisionTask.getHistory().getEvents().get(0).getWorkflowExecutionStartedEventAttributes(); In ReplayDecisionTaskHandler, we have special handling for sticky decision task that has a partial history. But we do not check the returned...

**What changed?** Add cluster ID into ringpop app **Why?** This is to prevent the membership merged from different cluster. **How did you test it?** local test **Potential risks** **Is hotfix...

**Is your feature request related to a problem? Please describe.** Currently, the behavior of delete workflow execution depends on the state of the task processing. (https://github.com/temporalio/temporal/blob/master/service/history/workflow/delete_manager.go#L117-L147) This is not a...

**What changed?** **Why?** **How did you test it?** **Potential risks** **Is hotfix candidate?**