grpcclients

func NewACBlobAccess ¶

func NewACBlobAccess(client grpc.ClientConnInterface, maximumMessageSizeBytes int) blobstore.BlobAccess

NewACBlobAccess creates a BlobAccess handle that relays any requests to a GRPC service that implements the remoteexecution.ActionCache service. That is the service that Bazel uses to access action results stored in the Action Cache.

func NewCASBlobAccess ¶

func NewCASBlobAccess(client grpc.ClientConnInterface, uuidGenerator util.UUIDGenerator, readChunkSize int) blobstore.BlobAccess

NewCASBlobAccess creates a BlobAccess handle that relays any requests to a GRPC service that implements the bytestream.ByteStream and remoteexecution.ContentAddressableStorage services. Those are the services that Bazel uses to access blobs stored in the Content Addressable Storage.

func NewFSACBlobAccess ¶

func NewFSACBlobAccess(client grpc.ClientConnInterface, maximumMessageSizeBytes int) blobstore.BlobAccess

NewFSACBlobAccess creates a BlobAccess that relays any requests to a gRPC server that implements the fsac.FileSystemAccessCache service.

This is a service that is specific to Buildbarn, used to store profiles of file system access patterns of build actions. These profiles can be used to perform readahead of objects stored in the action's input root.

func NewICASBlobAccess ¶

func NewICASBlobAccess(client grpc.ClientConnInterface, maximumMessageSizeBytes int) blobstore.BlobAccess

NewICASBlobAccess creates a BlobAccess that relays any requests to a gRPC server that implements the icas.IndirectContentAddressableStorage service. This is a service that is specific to Buildbarn, used to track references to objects stored in external corpora.

func NewISCCBlobAccess ¶

func NewISCCBlobAccess(client grpc.ClientConnInterface, maximumMessageSizeBytes int) blobstore.BlobAccess

NewISCCBlobAccess creates a BlobAccess that relays any requests to a gRPC server that implements the iscc.InitialSizeClassCacheClient service. This is a service that is specific to Buildbarn, used to let the scheduler store execution times of actions, so that it can make better predictions about which size class to pick during future invocations of similar actions.