Merge branch 'mablanch/823-action-cache-instance-name' into 'master'

Fix and improve instances support for remote-execution.

See merge request BuildStream/buildstream!1174

4 files changed, 47 insertions, 16 deletions

diff --git a/buildstream/_cas/casremote.py b/buildstream/_cas/casremote.py
index 56ba4c5d8..df1dd799c 100644
--- a/buildstream/_cas/casremote.py
+++ b/buildstream/_cas/casremote.py
@@ -31,7 +31,7 @@ class CASRemoteSpec(namedtuple('CASRemoteSpec', 'url push server_cert client_key
     #
     @staticmethod
     def _new_from_config_node(spec_node, basedir=None):
-        _yaml.node_validate(spec_node, ['url', 'push', 'server-cert', 'client-key', 'client-cert', 'instance_name'])
+        _yaml.node_validate(spec_node, ['url', 'push', 'server-cert', 'client-key', 'client-cert', 'instance-name'])
         url = _yaml.node_get(spec_node, str, 'url')
         push = _yaml.node_get(spec_node, bool, 'push', default_value=False)
         if not url:
@@ -39,7 +39,7 @@ class CASRemoteSpec(namedtuple('CASRemoteSpec', 'url push server_cert client_key
             raise LoadError(LoadErrorReason.INVALID_DATA,
                             "{}: empty artifact cache URL".format(provenance))
 
-        instance_name = _yaml.node_get(spec_node, str, 'server-cert', default_value=None)
+        instance_name = _yaml.node_get(spec_node, str, 'instance-name', default_value=None)
 
         server_cert = _yaml.node_get(spec_node, str, 'server-cert', default_value=None)
         if server_cert and basedir:
@@ -83,6 +83,7 @@ class CASRemote():
         self.spec = spec
         self._initialized = False
         self.channel = None
+        self.instance_name = None
         self.bytestream = None
         self.cas = None
         self.ref_storage = None
@@ -125,6 +126,8 @@ class CASRemote():
             else:
                 raise CASRemoteError("Unsupported URL: {}".format(self.spec.url))
 
+            self.instance_name = self.spec.instance_name or None
+
             self.bytestream = bytestream_pb2_grpc.ByteStreamStub(self.channel)
             self.cas = remote_execution_pb2_grpc.ContentAddressableStorageStub(self.channel)
             self.capabilities = remote_execution_pb2_grpc.CapabilitiesStub(self.channel)
@@ -133,6 +136,8 @@ class CASRemote():
             self.max_batch_total_size_bytes = _MAX_PAYLOAD_BYTES
             try:
                 request = remote_execution_pb2.GetCapabilitiesRequest()
+                if self.instance_name:
+                    request.instance_name = self.instance_name
                 response = self.capabilities.GetCapabilities(request)
                 server_max_batch_total_size_bytes = response.cache_capabilities.max_batch_total_size_bytes
                 if 0 < server_max_batch_total_size_bytes < self.max_batch_total_size_bytes:
@@ -146,6 +151,8 @@ class CASRemote():
             self.batch_read_supported = False
             try:
                 request = remote_execution_pb2.BatchReadBlobsRequest()
+                if self.instance_name:
+                    request.instance_name = self.instance_name
                 response = self.cas.BatchReadBlobs(request)
                 self.batch_read_supported = True
             except grpc.RpcError as e:
@@ -156,6 +163,8 @@ class CASRemote():
             self.batch_update_supported = False
             try:
                 request = remote_execution_pb2.BatchUpdateBlobsRequest()
+                if self.instance_name:
+                    request.instance_name = self.instance_name
                 response = self.cas.BatchUpdateBlobs(request)
                 self.batch_update_supported = True
             except grpc.RpcError as e:
@@ -224,6 +233,8 @@ class CASRemote():
         self.init()
 
         request = remote_execution_pb2.FindMissingBlobsRequest()
+        if self.instance_name:
+            request.instance_name = self.instance_name
         request.blob_digests.extend([digest])
 
         response = self.cas.FindMissingBlobs(request)
@@ -258,7 +269,13 @@ class CASRemote():
     #             Local Private Methods            #
     ################################################
     def _fetch_blob(self, digest, stream):
-        resource_name = '/'.join(['blobs', digest.hash, str(digest.size_bytes)])
+        if self.instance_name:
+            resource_name = '/'.join([self.instance_name, 'blobs',
+                                      digest.hash, str(digest.size_bytes)])
+        else:
+            resource_name = '/'.join(['blobs',
+                                      digest.hash, str(digest.size_bytes)])
+
         request = bytestream_pb2.ReadRequest()
         request.resource_name = resource_name
         request.read_offset = 0
@@ -269,8 +286,12 @@ class CASRemote():
         assert digest.size_bytes == os.fstat(stream.fileno()).st_size
 
     def _send_blob(self, digest, stream, u_uid=uuid.uuid4()):
-        resource_name = '/'.join(['uploads', str(u_uid), 'blobs',
-                                  digest.hash, str(digest.size_bytes)])
+        if self.instance_name:
+            resource_name = '/'.join([self.instance_name, 'uploads', str(u_uid), 'blobs',
+                                      digest.hash, str(digest.size_bytes)])
+        else:
+            resource_name = '/'.join(['uploads', str(u_uid), 'blobs',
+                                      digest.hash, str(digest.size_bytes)])
 
         def request_stream(resname, instream):
             offset = 0
@@ -304,6 +325,8 @@ class _CASBatchRead():
         self._remote = remote
         self._max_total_size_bytes = remote.max_batch_total_size_bytes
         self._request = remote_execution_pb2.BatchReadBlobsRequest()
+        if remote.instance_name:
+            self._request.instance_name = remote.instance_name
         self._size = 0
         self._sent = False
 
@@ -351,6 +374,8 @@ class _CASBatchUpdate():
         self._remote = remote
         self._max_total_size_bytes = remote.max_batch_total_size_bytes
         self._request = remote_execution_pb2.BatchUpdateBlobsRequest()
+        if remote.instance_name:
+            self._request.instance_name = remote.instance_name
         self._size = 0
         self._sent = False
 
diff --git a/buildstream/sandbox/_sandboxremote.py b/buildstream/sandbox/_sandboxremote.py
index 6a1f6f2f8..e9a4a7cb5 100644
--- a/buildstream/sandbox/_sandboxremote.py
+++ b/buildstream/sandbox/_sandboxremote.py
@@ -81,16 +81,18 @@ class SandboxRemote(Sandbox):
 
         if config.action_service:
             self.action_url = config.action_service['url']
+            self.action_instance = config.action_service.get('instance-name', None)
             self.action_credentials = grpc.ssl_channel_credentials(
                 root_certificates=action_certs.get('server-cert'),
                 private_key=action_certs.get('client-key'),
                 certificate_chain=action_certs.get('client-cert'))
         else:
             self.action_url = None
+            self.action_instance = None
             self.action_credentials = None
 
-        self.server_instance = config.exec_service.get('instance', None)
-        self.storage_instance = config.storage_service.get('instance', None)
+        self.exec_instance = config.exec_service.get('instance-name', None)
+        self.storage_instance = config.storage_service.get('instance-name', None)
 
         self.storage_remote_spec = CASRemoteSpec(self.storage_url, push=True,
                                                  server_cert=config.storage_service['server-cert'],
@@ -131,10 +133,10 @@ class SandboxRemote(Sandbox):
         remote_exec_storage_config = require_node(remote_config, 'storage-service')
         remote_exec_action_config = remote_config.get('action-cache-service', {})
 
-        _yaml.node_validate(remote_exec_service_config, ['url', 'instance'] + tls_keys)
-        _yaml.node_validate(remote_exec_storage_config, ['url', 'instance'] + tls_keys)
+        _yaml.node_validate(remote_exec_service_config, ['url', 'instance-name'] + tls_keys)
+        _yaml.node_validate(remote_exec_storage_config, ['url', 'instance-name'] + tls_keys)
         if remote_exec_action_config:
-            _yaml.node_validate(remote_exec_action_config, ['url'] + tls_keys)
+            _yaml.node_validate(remote_exec_action_config, ['url', 'instance-name'] + tls_keys)
         else:
             remote_config['action-service'] = None
 
@@ -182,7 +184,7 @@ class SandboxRemote(Sandbox):
 
         # Try to create a communication channel to the BuildGrid server.
         stub = remote_execution_pb2_grpc.ExecutionStub(channel)
-        request = remote_execution_pb2.ExecuteRequest(instance_name=self.server_instance,
+        request = remote_execution_pb2.ExecuteRequest(instance_name=self.exec_instance,
                                                       action_digest=action_digest,
                                                       skip_cache_lookup=False)
 
@@ -394,7 +396,8 @@ class SandboxRemote(Sandbox):
         elif url.scheme == 'https':
             channel = grpc.secure_channel('{}:{}'.format(url.hostname, url.port), self.action_credentials)
 
-        request = remote_execution_pb2.GetActionResultRequest(action_digest=action_digest)
+        request = remote_execution_pb2.GetActionResultRequest(instance_name=self.action_instance,
+                                                              action_digest=action_digest)
         stub = remote_execution_pb2_grpc.ActionCacheStub(channel)
         try:
             result = stub.GetActionResult(request)
diff --git a/doc/source/format_project.rst b/doc/source/format_project.rst
index f4dea333a..529816176 100644
--- a/doc/source/format_project.rst
+++ b/doc/source/format_project.rst
@@ -243,6 +243,7 @@ using the `remote-execution` option:
       instance-name: development-emea-1
     action-cache-service:
       url: http://bar.action.com:50052
+      instance-name: development-emea-1
 
 storage-service specifies a remote CAS store and the parameters are the
 same as those used to specify an :ref:`artifact server <artifacts>`.
diff --git a/doc/source/using_config.rst b/doc/source/using_config.rst
index 71ca64a4a..a7ee4dca9 100644
--- a/doc/source/using_config.rst
+++ b/doc/source/using_config.rst
@@ -122,13 +122,14 @@ configuration will be used as fallback.
       url: http://execution.fallback.example.com:50051
       instance-name: main
     storage-service:
-      url: https://storage.fallback.example.com:11002/
+      url: https://storage.fallback.example.com:11002
       server-cert: /keys/server.crt
       client-cert: /keys/client.crt
       client-key: /keys/client.key
       instance-name: main
     action-cache-service:
-      url: http://action.flalback.example.com:50052
+      url: http://cache.flalback.example.com:50052
+      instance-name: main
 
 2. Project override:
 
@@ -141,13 +142,14 @@ configuration will be used as fallback.
           url: http://execution.some_project.example.com:50051
           instance-name: main
         storage-service:
-          url: https://storage.some_project.example.com:11002/
+          url: https://storage.some_project.example.com:11002
           server-cert: /some_project_keys/server.crt
           client-cert: /some_project_keys/client.crt
           client-key: /some_project_keys/client.key
           instance-name: main
         action-cache-service:
-          url: http://action.some_project.example.com:50052
+          url: http://cache.some_project.example.com:50052
+          instance-name: main
 
 
 Strict build plan