-
-
Notifications
You must be signed in to change notification settings - Fork 108
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Storage write api - support default stream #226
Conversation
…t-stream Storage write api support default stream
+1 to this - @goccy is this repo actively maintained still? |
+1 actively need this as well as the default streaming method uses the "_default" stream and you can't use connections pool without that : I tried a workaround for this consisting of creating the "_default" write stream after the table but It seems like the API doesn't allow us to set the name of the stream ->
With the following code, the stream created have a randomly generated name. |
+1 i need this with go client. Here's my example Go code using the managedwriter package to request the bigquery emulator: ctx := context.Background()
grpcCtx, _ := context.WithTimeout(ctx, 1*time.Second)
conn, _ := grpc.DialContext(grpcCtx, "localhost:9060",
grpc.WithTransportCredentials(insecure.NewCredentials()),
grpc.WithIdleTimeout(1*time.Second),
)
client, err := managedwriter.NewClient(ctx, projectID,
option.WithGRPCConn(conn),
option.WithoutAuthentication(),
) // OK
pb := &v1.pb{}
descriptorProto, err := adapt.NormalizeDescriptor(pb.ProtoReflect().Descriptor())
tableName := fmt.Sprintf("projects/%s/datasets/%s/tables/%s", projectID, datasetID, "{table_name}")
stream, err := client.NewManagedStream(ctx,
managedwriter.WithDestinationTable(tableName),
managedwriter.WithSchemaDescriptor(descriptorProto),
managedwriter.WithAppendRowsCallOption(gax.WithTimeout(1*time.Second)),
) // return error
// rpc error: code = Unknown desc = failed to find stream from projects/{project}/datasets/{dataset}/tables/{table}/streams/_default When calling NewManagedStream() without specifying the stream type option, it internally invokes getWriteStream with the stream name argument IMO, the API should automatically create a _default stream for each table if it doesn't exist at the time of the call, or alternatively, prepare a _default stream for all tables during the initialization phase. Here's a quick fix I tried as an alternative, but I'm not sure if it's the right approach: araj-dev#1 |
+1 for default straem support |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Please add test cases for this fixing
# Conflicts: # server/storage_handler.go
@goccy added test case for default stream |
@goccy Could you please merge this PR into the main branch? I also need it. |
@MatanLevy I would like to check the CI results, so could you push an empty commit to trigger the CI ? |
@goccy pushed now an empty commit |
Thank you for your contribution !!! LGTM 👍 |
This pull request aims to support bigquery emulator for clients using storage write api with the default stream.
Two main requirements (based on google documentation) for default stream that are part of that pull request:
default stream creation
commited stream in AppendRows