Cluster not found #4501
Cluster not found #4501
Conversation
client/clientBean.go
Outdated
| @@ -184,7 +184,7 @@ func (h *clientBeanImpl) GetRemoteAdminClient(cluster string) (adminservice.Admi | |||
|
|
|||
| clusterInfo, clusterFound := h.clusterMetadata.GetAllClusterInfo()[cluster] | |||
| if !clusterFound { | |||
| return nil, &serviceerror.NotFound{ | |||
| return nil, &serviceerror.Unavailable{ | |||
There was a problem hiding this comment.
How do you differentiate real NOT_FOUND case?
There was a problem hiding this comment.
This not found err is translated to workflow not found on SDK side. This is kind of revert the change: https://github.com/temporalio/temporal/pull/3557/files#diff-0fb06c985229c9113e2a3fdd295d262429b58e003f73cabf860c432fc3529d47L194.
There was a problem hiding this comment.
We don't need to differentiate the not found. This is due to some server side misconfig and should not be a not found.
There was a problem hiding this comment.
Don't have the full context here but thought I'd drop by and mention that Unavailable error reaching a client counts against our SLA in cloud and is considered retryable vs. NotFound which should not be retried by default, I'm assuming this is desirable.
There was a problem hiding this comment.
I remember last time we changed this to NotFound so that standby task won't get stuck when remote cluster is not found.
There was a problem hiding this comment.
Maybe check-in a comment re. why it's not found? or you plan to fix it in a different way?
What changed?
Change cluster not found error to unavailable error
Why?
This is internal misconfigured and should return unavailable error.
How did you test it?
Potential risks
Is hotfix candidate?