Open
Description
dapr version: 1.13.5
Below picture is the topology of our services in Kubernetes. When we update Pod B, Pod A has lots of error message like below, seems like Pod A still use grpc connection to Pod B.
Pod B dapr sidecar has annotation graceful-shutdown-seconds, and container b has preStop probe.
dapr.io/graceful-shutdown-seconds: "25"
...
lifecycle:
preStop:
exec:
command:
- sh
- -c
- sleep 16
...
terminationGracePeriodSeconds: 30
rpc client unary retry. Err:rpc error: code = Unavailable desc = last connection error: connection error: desc = "transport: Error while dialing: dial tcp 10.110.0.2:50002: connect: connection refused" retry:2
Our question is how to avoid such problem when deployment rolling update?
Metadata
Metadata
Assignees
Labels
No labels