lukecwik commented on code in PR #25065:
URL: https://github.com/apache/beam/pull/25065#discussion_r1080787349
##########
model/fn-execution/src/main/proto/org/apache/beam/model/fn_execution/v1/beam_fn_api.proto:
##########
@@ -137,12 +138,64 @@ message InstructionResponse {
FinalizeBundleResponse finalize_bundle = 1004;
MonitoringInfosMetadataResponse monitoring_infos = 1005;
HarnessMonitoringInfosResponse harness_monitoring_infos = 1006;
+ SampleResponse sample = 1007;
// DEPRECATED
RegisterResponse register = 1000;
}
}
+// If supported, the `SampleRequest` will respond with a `SampleResponse` of
any
+// sampled elements for the PCollections within the given
+// ProcessBundleDescriptor. The SDK being queried must have the
+// "beam:protocol:data_sampling:v1" capability.
+message SampleRequest {
+ // Required. The id of the ProcessBundleDescriptor to sample. If empty,
+ // returns samples from all descriptors.
+ string process_bundle_descriptor_id = 1;
+
+ // Optional. The PCollections to sample from. If empty, returns all sampled
+ // PCollections from the given PBD. If both are empty, returns all samples.
+ repeated string pcollection_ids = 2;
+}
+
+// Contains sampled elements from all PCollections from a single
+// ProcessBundleDescriptor when the SDK is processing a bundle.
+message Samples {
+
+ // A sampled element. This is a proto message to allow for additional
+ // per-element metadata.
+ message Element {
+ // Required. Sampled raw bytes for an element. This is a
+ // single encoded element in the nested context.
+ bytes element = 1;
+
+ // FUTURE WORK: Capture lull detections and exceptions.
+ //
+ // Optional. Present if there was an exception
+ // processing the above element.
+ //
+ // LogEntry exception_entry = 2;
+ }
+
+ message ElementList {
+ // Required. The individual elements sampled from a PCollection.
+ repeated Element elements = 1;
+ }
+
+ // Required. Map from PCollection id to list of encoded elements using the
+ // associated Ccoder id for that PCollection.
+ map<string, ElementList> pcollections = 1;
+}
+
+// If supported, the `SampleResponse` will contain samples from all
+// ProcessBundleDescriptors.
+message SampleResponse {
+ // Required. Map from ProcessBundleDescriptor id to samples using that
+ // descriptor.
+ map<string, Samples> descriptors = 1;
Review Comment:
Not in any planned future and likely so far out of scope that many of these
concepts and designs would have to be modified.
Note that the runner can always generate PCollection ids that are unique
across pipelines as well if it were to come to that or we could evolve the API
to add this field when needed.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
