Is your feature request related to a problem? Please describe.
Cadence is moving towards supporting GRPC for RPC as opposed to the current/past implementation using Thrift.
Due to this effort they implemented a Thrift->GRPC compatibility layer which can support using GRPC instead of Thrift given the proper configuration.
Currently the Helm chart does not support exposing the GRPC port of the Cadence server frontend, but exposing it is a prerequisite for using GRPC.
The Helm chart should support this use-case.
Describe the solution you'd like to see
If I read it correctly, the frontend deployment (in addition to the exposed Thrift port 7933) would need to expose the GRPC port 7833here and then the corresponding frontend service could (in addition to the exposed Thrift port set by the values.yaml) expose the frontend GRPC port here and map it to the frontend pod 's GRPC port (plus any missed extras).
It is up for decision whether it is easier to expose all services' GRPC ports or limit it to the frontend service. The latter would be enough according to the Cadence client PR, trade-offs should be considered.
Is your feature request related to a problem? Please describe.
Cadence is moving towards supporting GRPC for RPC as opposed to the current/past implementation using Thrift. Due to this effort they implemented a Thrift->GRPC compatibility layer which can support using GRPC instead of Thrift given the proper configuration.
Currently the Helm chart does not support exposing the GRPC port of the Cadence server frontend, but exposing it is a prerequisite for using GRPC.
The Helm chart should support this use-case.
Describe the solution you'd like to see
If I read it correctly, the frontend deployment (in addition to the exposed Thrift port
7933
) would need to expose the GRPC port7833
here and then the corresponding frontend service could (in addition to the exposed Thrift port set by thevalues.yaml
) expose the frontend GRPC port here and map it to the frontend pod 's GRPC port (plus any missed extras).It is up for decision whether it is easier to expose all services' GRPC ports or limit it to the frontend service. The latter would be enough according to the Cadence client PR, trade-offs should be considered.
Describe alternatives you've considered
-
Additional context
Cadence client PR. Slack conversation about the feature request.