Martin Blanchard pushed to branch mablanch/83-executed-action-metadata at BuildGrid / buildgrid
Commits:
-
c720eeb9
by Martin Blanchard at 2018-10-23T13:35:50Z
-
e143fc8a
by Martin Blanchard at 2018-10-23T13:35:52Z
-
a0fe0456
by Martin Blanchard at 2018-10-23T13:36:13Z
-
2b7edb0d
by Martin Blanchard at 2018-10-23T13:36:15Z
-
8fe91833
by Finn at 2018-10-23T17:09:28Z
-
2a04f0ae
by Finn at 2018-10-24T08:05:59Z
-
cbeea07c
by Martin Blanchard at 2018-10-24T08:20:58Z
-
7aa62925
by Martin Blanchard at 2018-10-24T15:21:17Z
-
bfc2e34d
by Martin Blanchard at 2018-10-24T16:22:56Z
-
3cf2d15a
by Martin Blanchard at 2018-10-24T16:27:35Z
-
95b84d45
by Martin Blanchard at 2018-10-24T16:32:50Z
15 changed files:
- .pylintrc
- buildgrid/_app/bots/dummy.py
- buildgrid/_app/bots/host.py
- buildgrid/_app/cli.py
- buildgrid/_app/commands/cmd_operation.py
- buildgrid/server/bots/instance.py
- buildgrid/server/execution/instance.py
- buildgrid/server/job.py
- buildgrid/server/operations/instance.py
- buildgrid/server/scheduler.py
- buildgrid/utils.py
- setup.py
- tests/integration/bots_service.py
- tests/integration/execution_service.py
- tests/integration/operations_service.py
Changes:
| ... | ... | @@ -184,7 +184,8 @@ ignore-on-opaque-inference=yes |
| 184 | 184 |
# List of class names for which member attributes should not be checked (useful
|
| 185 | 185 |
# for classes with dynamically set attributes). This supports the use of
|
| 186 | 186 |
# qualified names.
|
| 187 |
-ignored-classes=google.protobuf.any_pb2.Any
|
|
| 187 |
+ignored-classes=google.protobuf.any_pb2.Any,
|
|
| 188 |
+ google.protobuf.timestamp_pb2.Timestamp
|
|
| 188 | 189 |
|
| 189 | 190 |
# List of module names for which member attributes should not be checked
|
| 190 | 191 |
# (useful for modules/projects where namespaces are manipulated during runtime
|
| ... | ... | @@ -17,16 +17,32 @@ import random |
| 17 | 17 |
import time
|
| 18 | 18 |
|
| 19 | 19 |
from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
|
| 20 |
+from buildgrid.utils import get_hostname
|
|
| 20 | 21 |
|
| 21 | 22 |
|
| 22 | 23 |
def work_dummy(context, lease):
|
| 23 | 24 |
""" Just returns lease after some random time
|
| 24 | 25 |
"""
|
| 26 |
+ action_result = remote_execution_pb2.ActionResult()
|
|
| 27 |
+ |
|
| 25 | 28 |
lease.result.Clear()
|
| 26 | 29 |
|
| 27 |
- time.sleep(random.randint(1, 5))
|
|
| 30 |
+ action_result.execution_metadata.worker = get_hostname()
|
|
| 28 | 31 |
|
| 29 |
- action_result = remote_execution_pb2.ActionResult()
|
|
| 32 |
+ # Simulation input-downloading phase:
|
|
| 33 |
+ action_result.execution_metadata.input_fetch_start_timestamp.GetCurrentTime()
|
|
| 34 |
+ time.sleep(random.random())
|
|
| 35 |
+ action_result.execution_metadata.input_fetch_completed_timestamp.GetCurrentTime()
|
|
| 36 |
+ |
|
| 37 |
+ # Simulation execution phase:
|
|
| 38 |
+ action_result.execution_metadata.execution_start_timestamp.GetCurrentTime()
|
|
| 39 |
+ time.sleep(random.random())
|
|
| 40 |
+ action_result.execution_metadata.execution_completed_timestamp.GetCurrentTime()
|
|
| 41 |
+ |
|
| 42 |
+ # Simulation output-uploading phase:
|
|
| 43 |
+ action_result.execution_metadata.output_upload_start_timestamp.GetCurrentTime()
|
|
| 44 |
+ time.sleep(random.random())
|
|
| 45 |
+ action_result.execution_metadata.output_upload_completed_timestamp.GetCurrentTime()
|
|
| 30 | 46 |
|
| 31 | 47 |
lease.result.Pack(action_result)
|
| 32 | 48 |
|
| ... | ... | @@ -19,7 +19,7 @@ import tempfile |
| 19 | 19 |
|
| 20 | 20 |
from buildgrid.client.cas import download, upload
|
| 21 | 21 |
from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
|
| 22 |
-from buildgrid.utils import output_file_maker, output_directory_maker
|
|
| 22 |
+from buildgrid.utils import get_hostname, output_file_maker, output_directory_maker
|
|
| 23 | 23 |
|
| 24 | 24 |
|
| 25 | 25 |
def work_host_tools(context, lease):
|
| ... | ... | @@ -29,10 +29,13 @@ def work_host_tools(context, lease): |
| 29 | 29 |
logger = context.logger
|
| 30 | 30 |
|
| 31 | 31 |
action_digest = remote_execution_pb2.Digest()
|
| 32 |
+ action_result = remote_execution_pb2.ActionResult()
|
|
| 32 | 33 |
|
| 33 | 34 |
lease.payload.Unpack(action_digest)
|
| 34 | 35 |
lease.result.Clear()
|
| 35 | 36 |
|
| 37 |
+ action_result.execution_metadata.worker = get_hostname()
|
|
| 38 |
+ |
|
| 36 | 39 |
with tempfile.TemporaryDirectory() as temp_directory:
|
| 37 | 40 |
with download(context.cas_channel, instance=instance_name) as downloader:
|
| 38 | 41 |
action = downloader.get_message(action_digest,
|
| ... | ... | @@ -43,8 +46,12 @@ def work_host_tools(context, lease): |
| 43 | 46 |
command = downloader.get_message(action.command_digest,
|
| 44 | 47 |
remote_execution_pb2.Command())
|
| 45 | 48 |
|
| 49 |
+ action_result.execution_metadata.input_fetch_start_timestamp.GetCurrentTime()
|
|
| 50 |
+ |
|
| 46 | 51 |
downloader.download_directory(action.input_root_digest, temp_directory)
|
| 47 | 52 |
|
| 53 |
+ action_result.execution_metadata.input_fetch_completed_timestamp.GetCurrentTime()
|
|
| 54 |
+ |
|
| 48 | 55 |
environment = os.environ.copy()
|
| 49 | 56 |
for variable in command.environment_variables:
|
| 50 | 57 |
if variable.name not in ['PATH', 'PWD']:
|
| ... | ... | @@ -70,6 +77,8 @@ def work_host_tools(context, lease): |
| 70 | 77 |
|
| 71 | 78 |
logger.debug(' '.join(command_line))
|
| 72 | 79 |
|
| 80 |
+ action_result.execution_metadata.execution_start_timestamp.GetCurrentTime()
|
|
| 81 |
+ |
|
| 73 | 82 |
process = subprocess.Popen(command_line,
|
| 74 | 83 |
cwd=working_directory,
|
| 75 | 84 |
env=environment,
|
| ... | ... | @@ -80,7 +89,8 @@ def work_host_tools(context, lease): |
| 80 | 89 |
stdout, stderr = process.communicate()
|
| 81 | 90 |
returncode = process.returncode
|
| 82 | 91 |
|
| 83 |
- action_result = remote_execution_pb2.ActionResult()
|
|
| 92 |
+ action_result.execution_metadata.execution_completed_timestamp.GetCurrentTime()
|
|
| 93 |
+ |
|
| 84 | 94 |
# TODO: Upload to CAS or output RAW
|
| 85 | 95 |
# For now, just pass raw
|
| 86 | 96 |
# https://gitlab.com/BuildGrid/buildgrid/issues/90
|
| ... | ... | @@ -92,6 +102,8 @@ def work_host_tools(context, lease): |
| 92 | 102 |
logger.debug("Command stdout: [{}]".format(stdout))
|
| 93 | 103 |
logger.debug("Command exit code: [{}]".format(returncode))
|
| 94 | 104 |
|
| 105 |
+ action_result.execution_metadata.output_upload_start_timestamp.GetCurrentTime()
|
|
| 106 |
+ |
|
| 95 | 107 |
with upload(context.cas_channel, instance=instance_name) as uploader:
|
| 96 | 108 |
output_files, output_directories = [], []
|
| 97 | 109 |
|
| ... | ... | @@ -121,6 +133,8 @@ def work_host_tools(context, lease): |
| 121 | 133 |
|
| 122 | 134 |
action_result.output_directories.extend(output_directories)
|
| 123 | 135 |
|
| 136 |
+ action_result.execution_metadata.output_upload_completed_timestamp.GetCurrentTime()
|
|
| 137 |
+ |
|
| 124 | 138 |
lease.result.Pack(action_result)
|
| 125 | 139 |
|
| 126 | 140 |
return lease
|
| ... | ... | @@ -26,7 +26,6 @@ import logging |
| 26 | 26 |
|
| 27 | 27 |
import click
|
| 28 | 28 |
import grpc
|
| 29 |
-from xdg import XDG_CACHE_HOME, XDG_CONFIG_HOME, XDG_DATA_HOME
|
|
| 30 | 29 |
|
| 31 | 30 |
from buildgrid.utils import read_file
|
| 32 | 31 |
|
| ... | ... | @@ -42,41 +41,21 @@ class Context: |
| 42 | 41 |
|
| 43 | 42 |
self.user_home = os.getcwd()
|
| 44 | 43 |
|
| 45 |
- self.cache_home = os.path.join(XDG_CACHE_HOME, 'buildgrid')
|
|
| 46 |
- self.config_home = os.path.join(XDG_CONFIG_HOME, 'buildgrid')
|
|
| 47 |
- self.data_home = os.path.join(XDG_DATA_HOME, 'buildgrid')
|
|
| 48 |
- |
|
| 49 |
- def load_client_credentials(self, client_key=None, client_cert=None,
|
|
| 50 |
- server_cert=None, use_default_client_keys=False):
|
|
| 44 |
+ def load_client_credentials(self, client_key=None, client_cert=None, server_cert=None):
|
|
| 51 | 45 |
"""Looks-up and loads TLS client gRPC credentials.
|
| 52 | 46 |
|
| 53 | 47 |
Args:
|
| 54 | 48 |
client_key(str): root certificate file path.
|
| 55 | 49 |
client_cert(str): private key file path.
|
| 56 | 50 |
server_cert(str): certificate chain file path.
|
| 57 |
- use_default_client_keys(bool, optional): whether or not to try
|
|
| 58 |
- loading client keys from default location. Defaults to False.
|
|
| 59 | 51 |
|
| 60 | 52 |
Returns:
|
| 61 | 53 |
:obj:`ChannelCredentials`: The credentials for use for a
|
| 62 | 54 |
TLS-encrypted gRPC client channel.
|
| 63 | 55 |
"""
|
| 64 |
- if not client_key or not os.path.exists(client_key):
|
|
| 65 |
- if use_default_client_keys:
|
|
| 66 |
- client_key = os.path.join(self.config_home, 'client.key')
|
|
| 67 |
- else:
|
|
| 68 |
- client_key = None
|
|
| 69 |
- |
|
| 70 |
- if not client_cert or not os.path.exists(client_cert):
|
|
| 71 |
- if use_default_client_keys:
|
|
| 72 |
- client_cert = os.path.join(self.config_home, 'client.crt')
|
|
| 73 |
- else:
|
|
| 74 |
- client_cert = None
|
|
| 75 | 56 |
|
| 76 | 57 |
if not server_cert or not os.path.exists(server_cert):
|
| 77 |
- server_cert = os.path.join(self.config_home, 'server.crt')
|
|
| 78 |
- if not os.path.exists(server_cert):
|
|
| 79 |
- return None
|
|
| 58 |
+ return None
|
|
| 80 | 59 |
|
| 81 | 60 |
server_cert_pem = read_file(server_cert)
|
| 82 | 61 |
if client_key and os.path.exists(client_key):
|
| ... | ... | @@ -100,8 +79,7 @@ class Context: |
| 100 | 79 |
|
| 101 | 80 |
return credentials
|
| 102 | 81 |
|
| 103 |
- def load_server_credentials(self, server_key=None, server_cert=None,
|
|
| 104 |
- client_certs=None, use_default_client_certs=False):
|
|
| 82 |
+ def load_server_credentials(self, server_key=None, server_cert=None, client_certs=None):
|
|
| 105 | 83 |
"""Looks-up and loads TLS server gRPC credentials.
|
| 106 | 84 |
|
| 107 | 85 |
Every private and public keys are expected to be PEM-encoded.
|
| ... | ... | @@ -110,29 +88,16 @@ class Context: |
| 110 | 88 |
server_key(str): private server key file path.
|
| 111 | 89 |
server_cert(str): public server certificate file path.
|
| 112 | 90 |
client_certs(str): public client certificates file path.
|
| 113 |
- use_default_client_certs(bool, optional): whether or not to try
|
|
| 114 |
- loading public client certificates from default location.
|
|
| 115 |
- Defaults to False.
|
|
| 116 | 91 |
|
| 117 | 92 |
Returns:
|
| 118 | 93 |
:obj:`ServerCredentials`: The credentials for use for a
|
| 119 | 94 |
TLS-encrypted gRPC server channel.
|
| 120 | 95 |
"""
|
| 121 | 96 |
if not server_key or not os.path.exists(server_key):
|
| 122 |
- server_key = os.path.join(self.config_home, 'server.key')
|
|
| 123 |
- if not os.path.exists(server_key):
|
|
| 124 |
- return None
|
|
| 97 |
+ return None
|
|
| 125 | 98 |
|
| 126 | 99 |
if not server_cert or not os.path.exists(server_cert):
|
| 127 |
- server_cert = os.path.join(self.config_home, 'server.crt')
|
|
| 128 |
- if not os.path.exists(server_cert):
|
|
| 129 |
- return None
|
|
| 130 |
- |
|
| 131 |
- if not client_certs or not os.path.exists(client_certs):
|
|
| 132 |
- if use_default_client_certs:
|
|
| 133 |
- client_certs = os.path.join(self.config_home, 'client.crt')
|
|
| 134 |
- else:
|
|
| 135 |
- client_certs = None
|
|
| 100 |
+ return None
|
|
| 136 | 101 |
|
| 137 | 102 |
server_key_pem = read_file(server_key)
|
| 138 | 103 |
server_cert_pem = read_file(server_cert)
|
| ... | ... | @@ -29,6 +29,7 @@ import grpc |
| 29 | 29 |
|
| 30 | 30 |
from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2, remote_execution_pb2_grpc
|
| 31 | 31 |
from buildgrid._protos.google.longrunning import operations_pb2, operations_pb2_grpc
|
| 32 |
+from buildgrid._protos.google.rpc import code_pb2
|
|
| 32 | 33 |
|
| 33 | 34 |
from ..cli import pass_context
|
| 34 | 35 |
|
| ... | ... | @@ -65,35 +66,98 @@ def cli(context, remote, instance_name, client_key, client_cert, server_cert): |
| 65 | 66 |
context.logger.debug("Starting for remote {}".format(context.remote))
|
| 66 | 67 |
|
| 67 | 68 |
|
| 69 |
+def _print_operation_state(operation, print_details=False):
|
|
| 70 |
+ metadata = remote_execution_pb2.ExecuteOperationMetadata()
|
|
| 71 |
+ # The metadata is expected to be an ExecuteOperationMetadata message:
|
|
| 72 |
+ assert operation.metadata.Is(metadata.DESCRIPTOR)
|
|
| 73 |
+ operation.metadata.Unpack(metadata)
|
|
| 74 |
+ |
|
| 75 |
+ stage_name = remote_execution_pb2.ExecuteOperationMetadata.Stage.Name(
|
|
| 76 |
+ metadata.stage).upper()
|
|
| 77 |
+ |
|
| 78 |
+ if not operation.done:
|
|
| 79 |
+ if stage_name == 'CACHE_CHECK':
|
|
| 80 |
+ click.echo('CacheCheck: {}: Querying action-cache (stage={})'
|
|
| 81 |
+ .format(operation.name, metadata.stage))
|
|
| 82 |
+ elif stage_name == 'QUEUED':
|
|
| 83 |
+ click.echo('Queued: {}: Waiting for execution (stage={})'
|
|
| 84 |
+ .format(operation.name, metadata.stage))
|
|
| 85 |
+ elif stage_name == 'EXECUTING':
|
|
| 86 |
+ click.echo('Executing: {}: Currently running (stage={})'
|
|
| 87 |
+ .format(operation.name, metadata.stage))
|
|
| 88 |
+ else:
|
|
| 89 |
+ click.echo('Error: {}: In an invalid state (stage={})'
|
|
| 90 |
+ .format(operation.name, metadata.stage), err=True)
|
|
| 91 |
+ return
|
|
| 92 |
+ |
|
| 93 |
+ response = remote_execution_pb2.ExecuteResponse()
|
|
| 94 |
+ # The response is expected to be an ExecutionResponse message:
|
|
| 95 |
+ assert operation.response.Is(response.DESCRIPTOR)
|
|
| 96 |
+ operation.response.Unpack(response)
|
|
| 97 |
+ |
|
| 98 |
+ if response.status.code != code_pb2.OK:
|
|
| 99 |
+ click.echo('Failure: {}: {} (code={})'
|
|
| 100 |
+ .format(operation.name, response.status.message, response.status.code))
|
|
| 101 |
+ else:
|
|
| 102 |
+ if response.result.exit_code != 0:
|
|
| 103 |
+ click.echo('Success: {}: Completed with failure (stage={}, exit_code={})'
|
|
| 104 |
+ .format(operation.name, metadata.stage, response.result.exit_code))
|
|
| 105 |
+ else:
|
|
| 106 |
+ click.echo('Success: {}: Completed succesfully (stage={}, exit_code={})'
|
|
| 107 |
+ .format(operation.name, metadata.stage, response.result.exit_code))
|
|
| 108 |
+ |
|
| 109 |
+ if print_details:
|
|
| 110 |
+ metadata = response.result.execution_metadata
|
|
| 111 |
+ click.echo(' worker={}'.format(metadata.worker))
|
|
| 112 |
+ |
|
| 113 |
+ queued = metadata.queued_timestamp.ToDatetime()
|
|
| 114 |
+ click.echo(' queued_at={}'.format(queued))
|
|
| 115 |
+ |
|
| 116 |
+ worker_start = metadata.worker_start_timestamp.ToDatetime()
|
|
| 117 |
+ worker_completed = metadata.worker_completed_timestamp.ToDatetime()
|
|
| 118 |
+ click.echo(' work_duration={}'.format(worker_completed - worker_start))
|
|
| 119 |
+ |
|
| 120 |
+ fetch_start = metadata.input_fetch_start_timestamp.ToDatetime()
|
|
| 121 |
+ fetch_completed = metadata.input_fetch_completed_timestamp.ToDatetime()
|
|
| 122 |
+ click.echo(' fetch_duration={}'.format(fetch_completed - fetch_start))
|
|
| 123 |
+ |
|
| 124 |
+ execution_start = metadata.execution_start_timestamp.ToDatetime()
|
|
| 125 |
+ execution_completed = metadata.execution_completed_timestamp.ToDatetime()
|
|
| 126 |
+ click.echo(' exection_duration={}'.format(execution_completed - execution_start))
|
|
| 127 |
+ |
|
| 128 |
+ upload_start = metadata.output_upload_start_timestamp.ToDatetime()
|
|
| 129 |
+ upload_completed = metadata.output_upload_completed_timestamp.ToDatetime()
|
|
| 130 |
+ click.echo(' upload_duration={}'.format(upload_completed - upload_start))
|
|
| 131 |
+ |
|
| 132 |
+ click.echo(' total_duration={}'.format(worker_completed - queued))
|
|
| 133 |
+ |
|
| 134 |
+ |
|
| 68 | 135 |
@cli.command('status', short_help="Get the status of an operation.")
|
| 69 | 136 |
@click.argument('operation-name', nargs=1, type=click.STRING, required=True)
|
| 70 | 137 |
@pass_context
|
| 71 | 138 |
def status(context, operation_name):
|
| 72 |
- context.logger.info("Getting operation status...")
|
|
| 73 | 139 |
stub = operations_pb2_grpc.OperationsStub(context.channel)
|
| 74 |
- |
|
| 75 | 140 |
request = operations_pb2.GetOperationRequest(name=operation_name)
|
| 76 | 141 |
|
| 77 |
- response = stub.GetOperation(request)
|
|
| 78 |
- context.logger.info(response)
|
|
| 142 |
+ operation = stub.GetOperation(request)
|
|
| 143 |
+ |
|
| 144 |
+ _print_operation_state(operation, print_details=True)
|
|
| 79 | 145 |
|
| 80 | 146 |
|
| 81 | 147 |
@cli.command('list', short_help="List operations.")
|
| 82 | 148 |
@pass_context
|
| 83 | 149 |
def lists(context):
|
| 84 |
- context.logger.info("Getting list of operations")
|
|
| 85 | 150 |
stub = operations_pb2_grpc.OperationsStub(context.channel)
|
| 86 |
- |
|
| 87 | 151 |
request = operations_pb2.ListOperationsRequest(name=context.instance_name)
|
| 88 | 152 |
|
| 89 | 153 |
response = stub.ListOperations(request)
|
| 90 | 154 |
|
| 91 | 155 |
if not response.operations:
|
| 92 |
- context.logger.warning("No operations to list")
|
|
| 156 |
+ click.echo('Error: No operations to list', err=True)
|
|
| 93 | 157 |
return
|
| 94 | 158 |
|
| 95 |
- for op in response.operations:
|
|
| 96 |
- context.logger.info(op)
|
|
| 159 |
+ for operation in response.operations:
|
|
| 160 |
+ _print_operation_state(operation)
|
|
| 97 | 161 |
|
| 98 | 162 |
|
| 99 | 163 |
@cli.command('wait', short_help="Streams an operation until it is complete.")
|
| ... | ... | @@ -103,7 +167,10 @@ def wait(context, operation_name): |
| 103 | 167 |
stub = remote_execution_pb2_grpc.ExecutionStub(context.channel)
|
| 104 | 168 |
request = remote_execution_pb2.WaitExecutionRequest(name=operation_name)
|
| 105 | 169 |
|
| 106 |
- response = stub.WaitExecution(request)
|
|
| 170 |
+ operation_iterator = stub.WaitExecution(request)
|
|
| 107 | 171 |
|
| 108 |
- for stream in response:
|
|
| 109 |
- context.logger.info(stream)
|
|
| 172 |
+ for operation in operation_iterator:
|
|
| 173 |
+ if operation.done:
|
|
| 174 |
+ _print_operation_state(operation, print_details=True)
|
|
| 175 |
+ else:
|
|
| 176 |
+ _print_operation_state(operation)
|
| ... | ... | @@ -66,10 +66,10 @@ class BotsInterface: |
| 66 | 66 |
self._bot_sessions[name] = bot_session
|
| 67 | 67 |
self.logger.info("Created bot session name=[{}] with bot_id=[{}]".format(name, bot_id))
|
| 68 | 68 |
|
| 69 |
- # For now, one lease at a time.
|
|
| 70 |
- lease = self._scheduler.create_lease()
|
|
| 71 |
- if lease:
|
|
| 72 |
- bot_session.leases.extend([lease])
|
|
| 69 |
+ # TODO: Send worker capabilities to the scheduler!
|
|
| 70 |
+ leases = self._scheduler.request_job_leases({})
|
|
| 71 |
+ if leases:
|
|
| 72 |
+ bot_session.leases.extend(leases)
|
|
| 73 | 73 |
|
| 74 | 74 |
return bot_session
|
| 75 | 75 |
|
| ... | ... | @@ -85,11 +85,11 @@ class BotsInterface: |
| 85 | 85 |
del bot_session.leases[:]
|
| 86 | 86 |
bot_session.leases.extend(leases)
|
| 87 | 87 |
|
| 88 |
- # For now, one lease at a time
|
|
| 88 |
+ # TODO: Send worker capabilities to the scheduler!
|
|
| 89 | 89 |
if not bot_session.leases:
|
| 90 |
- lease = self._scheduler.create_lease()
|
|
| 91 |
- if lease:
|
|
| 92 |
- bot_session.leases.extend([lease])
|
|
| 90 |
+ leases = self._scheduler.request_job_leases({})
|
|
| 91 |
+ if leases:
|
|
| 92 |
+ bot_session.leases.extend(leases)
|
|
| 93 | 93 |
|
| 94 | 94 |
self._bot_sessions[name] = bot_session
|
| 95 | 95 |
return bot_session
|
| ... | ... | @@ -109,7 +109,8 @@ class BotsInterface: |
| 109 | 109 |
if server_state == LeaseState.PENDING:
|
| 110 | 110 |
|
| 111 | 111 |
if client_state == LeaseState.ACTIVE:
|
| 112 |
- self._scheduler.update_job_lease_state(client_lease.id, client_lease.state)
|
|
| 112 |
+ self._scheduler.update_job_lease_state(client_lease.id,
|
|
| 113 |
+ LeaseState.ACTIVE)
|
|
| 113 | 114 |
elif client_state == LeaseState.COMPLETED:
|
| 114 | 115 |
# TODO: Lease was rejected
|
| 115 | 116 |
raise NotImplementedError("'Not Accepted' is unsupported")
|
| ... | ... | @@ -122,8 +123,10 @@ class BotsInterface: |
| 122 | 123 |
pass
|
| 123 | 124 |
|
| 124 | 125 |
elif client_state == LeaseState.COMPLETED:
|
| 125 |
- self._scheduler.update_job_lease_state(client_lease.id, client_lease.state)
|
|
| 126 |
- self._scheduler.job_complete(client_lease.id, client_lease.result, client_lease.status)
|
|
| 126 |
+ self._scheduler.update_job_lease_state(client_lease.id,
|
|
| 127 |
+ LeaseState.COMPLETED,
|
|
| 128 |
+ lease_status=client_lease.status,
|
|
| 129 |
+ lease_result=client_lease.result)
|
|
| 127 | 130 |
return None
|
| 128 | 131 |
|
| 129 | 132 |
else:
|
| ... | ... | @@ -48,12 +48,15 @@ class ExecutionInstance: |
| 48 | 48 |
if not action:
|
| 49 | 49 |
raise FailedPreconditionError("Could not get action from storage.")
|
| 50 | 50 |
|
| 51 |
- job = Job(action_digest, action.do_not_cache, message_queue)
|
|
| 51 |
+ job = Job(action, action_digest)
|
|
| 52 |
+ if message_queue is not None:
|
|
| 53 |
+ job.register_client(message_queue)
|
|
| 54 |
+ |
|
| 52 | 55 |
self.logger.info("Operation name: [{}]".format(job.name))
|
| 53 | 56 |
|
| 54 |
- self._scheduler.append_job(job, skip_cache_lookup)
|
|
| 57 |
+ self._scheduler.queue_job(job, skip_cache_lookup)
|
|
| 55 | 58 |
|
| 56 |
- return job.get_operation()
|
|
| 59 |
+ return job.operation
|
|
| 57 | 60 |
|
| 58 | 61 |
def register_message_client(self, name, queue):
|
| 59 | 62 |
try:
|
| 1 |
-# Copyright (C) 2018 Codethink Limited
|
|
| 1 |
+# Copyright (C) 2018 Bloomberg LP
|
|
| 2 | 2 |
#
|
| 3 | 3 |
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 4 | 4 |
# you may not use this file except in compliance with the License.
|
| ... | ... | @@ -11,14 +11,14 @@ |
| 11 | 11 |
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 12 | 12 |
# See the License for the specific language governing permissions and
|
| 13 | 13 |
# limitations under the License.
|
| 14 |
-#
|
|
| 15 |
-# Authors:
|
|
| 16 |
-# Finn Ball <finn ball codethink co uk>
|
|
| 14 |
+ |
|
| 17 | 15 |
|
| 18 | 16 |
import logging
|
| 19 | 17 |
import uuid
|
| 20 | 18 |
from enum import Enum
|
| 21 | 19 |
|
| 20 |
+from google.protobuf import timestamp_pb2
|
|
| 21 |
+ |
|
| 22 | 22 |
from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
|
| 23 | 23 |
from buildgrid._protos.google.devtools.remoteworkers.v1test2 import bots_pb2
|
| 24 | 24 |
from buildgrid._protos.google.longrunning import operations_pb2
|
| ... | ... | @@ -52,79 +52,190 @@ class LeaseState(Enum): |
| 52 | 52 |
|
| 53 | 53 |
class Job:
|
| 54 | 54 |
|
| 55 |
- def __init__(self, action_digest, do_not_cache=False, message_queue=None):
|
|
| 56 |
- self.lease = None
|
|
| 55 |
+ def __init__(self, action, action_digest):
|
|
| 57 | 56 |
self.logger = logging.getLogger(__name__)
|
| 58 |
- self.n_tries = 0
|
|
| 59 |
- self.result = None
|
|
| 60 |
- self.result_cached = False
|
|
| 61 | 57 |
|
| 62 |
- self._action_digest = action_digest
|
|
| 63 |
- self._do_not_cache = do_not_cache
|
|
| 64 |
- self._execute_stage = OperationStage.UNKNOWN
|
|
| 65 | 58 |
self._name = str(uuid.uuid4())
|
| 66 |
- self._operation = operations_pb2.Operation(name=self._name)
|
|
| 67 |
- self._operation_update_queues = []
|
|
| 59 |
+ self._action = remote_execution_pb2.Action()
|
|
| 60 |
+ self._operation = operations_pb2.Operation()
|
|
| 61 |
+ self._lease = None
|
|
| 62 |
+ |
|
| 63 |
+ self.__execute_response = None
|
|
| 64 |
+ self.__operation_metadata = remote_execution_pb2.ExecuteOperationMetadata()
|
|
| 65 |
+ self.__queued_timestamp = timestamp_pb2.Timestamp()
|
|
| 66 |
+ self.__worker_start_timestamp = timestamp_pb2.Timestamp()
|
|
| 67 |
+ self.__worker_completed_timestamp = timestamp_pb2.Timestamp()
|
|
| 68 |
+ |
|
| 69 |
+ self.__operation_metadata.action_digest.CopyFrom(action_digest)
|
|
| 70 |
+ self.__operation_metadata.stage = OperationStage.UNKNOWN.value
|
|
| 68 | 71 |
|
| 69 |
- if message_queue is not None:
|
|
| 70 |
- self.register_client(message_queue)
|
|
| 72 |
+ self._action.CopyFrom(action)
|
|
| 73 |
+ self._do_not_cache = self._action.do_not_cache
|
|
| 74 |
+ self._operation_update_queues = []
|
|
| 75 |
+ self._operation.name = self._name
|
|
| 76 |
+ self._operation.done = False
|
|
| 77 |
+ self._n_tries = 0
|
|
| 71 | 78 |
|
| 72 | 79 |
@property
|
| 73 | 80 |
def name(self):
|
| 74 | 81 |
return self._name
|
| 75 | 82 |
|
| 83 |
+ @property
|
|
| 84 |
+ def do_not_cache(self):
|
|
| 85 |
+ return self._do_not_cache
|
|
| 86 |
+ |
|
| 87 |
+ @property
|
|
| 88 |
+ def action(self):
|
|
| 89 |
+ return self._action
|
|
| 90 |
+ |
|
| 76 | 91 |
@property
|
| 77 | 92 |
def action_digest(self):
|
| 78 |
- return self._action_digest
|
|
| 93 |
+ return self.__operation_metadata.action_digest
|
|
| 79 | 94 |
|
| 80 | 95 |
@property
|
| 81 |
- def do_not_cache(self):
|
|
| 82 |
- return self._do_not_cache
|
|
| 96 |
+ def action_result(self):
|
|
| 97 |
+ if self.__execute_response is not None:
|
|
| 98 |
+ return self.__execute_response.result
|
|
| 99 |
+ else:
|
|
| 100 |
+ return None
|
|
| 101 |
+ |
|
| 102 |
+ @property
|
|
| 103 |
+ def operation(self):
|
|
| 104 |
+ return self._operation
|
|
| 105 |
+ |
|
| 106 |
+ @property
|
|
| 107 |
+ def operation_stage(self):
|
|
| 108 |
+ return OperationStage(self.__operation_metadata.state)
|
|
| 109 |
+ |
|
| 110 |
+ @property
|
|
| 111 |
+ def lease(self):
|
|
| 112 |
+ return self._lease
|
|
| 113 |
+ |
|
| 114 |
+ @property
|
|
| 115 |
+ def lease_state(self):
|
|
| 116 |
+ if self._lease is not None:
|
|
| 117 |
+ return LeaseState(self._lease.state)
|
|
| 118 |
+ else:
|
|
| 119 |
+ return None
|
|
| 83 | 120 |
|
| 84 |
- def check_job_finished(self):
|
|
| 85 |
- if not self._operation_update_queues:
|
|
| 86 |
- return self._operation.done
|
|
| 87 |
- return False
|
|
| 121 |
+ @property
|
|
| 122 |
+ def n_tries(self):
|
|
| 123 |
+ return self._n_tries
|
|
| 124 |
+ |
|
| 125 |
+ @property
|
|
| 126 |
+ def n_clients(self):
|
|
| 127 |
+ return len(self._operation_update_queues)
|
|
| 88 | 128 |
|
| 89 | 129 |
def register_client(self, queue):
|
| 130 |
+ """Subscribes to the job's :class:`Operation` stage change events.
|
|
| 131 |
+ |
|
| 132 |
+ Args:
|
|
| 133 |
+ queue (queue.Queue): the event queue to register.
|
|
| 134 |
+ """
|
|
| 90 | 135 |
self._operation_update_queues.append(queue)
|
| 91 |
- queue.put(self.get_operation())
|
|
| 136 |
+ queue.put(self._operation)
|
|
| 92 | 137 |
|
| 93 | 138 |
def unregister_client(self, queue):
|
| 139 |
+ """Unsubscribes to the job's :class:`Operation` stage change events.
|
|
| 140 |
+ |
|
| 141 |
+ Args:
|
|
| 142 |
+ queue (queue.Queue): the event queue to unregister.
|
|
| 143 |
+ """
|
|
| 94 | 144 |
self._operation_update_queues.remove(queue)
|
| 95 | 145 |
|
| 96 |
- def get_operation(self):
|
|
| 97 |
- self._operation.metadata.Pack(self.get_operation_meta())
|
|
| 98 |
- if self.result is not None:
|
|
| 99 |
- self._operation.done = True
|
|
| 100 |
- response = remote_execution_pb2.ExecuteResponse(result=self.result,
|
|
| 101 |
- cached_result=self.result_cached)
|
|
| 146 |
+ def set_cached_result(self, action_result):
|
|
| 147 |
+ """Allows specifying an action result form the action cache for the job.
|
|
| 148 |
+ """
|
|
| 149 |
+ self.__execute_response = remote_execution_pb2.ExecuteResponse()
|
|
| 150 |
+ self.__execute_response.result.CopyFrom(action_result)
|
|
| 151 |
+ self.__execute_response.cached_result = True
|
|
| 102 | 152 |
|
| 103 |
- if not self.result_cached:
|
|
| 104 |
- response.status.CopyFrom(self.lease.status)
|
|
| 153 |
+ def create_lease(self):
|
|
| 154 |
+ """Emits a new :class:`Lease` for the job.
|
|
| 105 | 155 |
|
| 106 |
- self._operation.response.Pack(response)
|
|
| 156 |
+ Only one :class:`Lease` can be emitted for a given job. This method
|
|
| 157 |
+ should only be used once, any furhter calls are ignored.
|
|
| 158 |
+ """
|
|
| 159 |
+ if self._lease is not None:
|
|
| 160 |
+ return None
|
|
| 107 | 161 |
|
| 108 |
- return self._operation
|
|
| 162 |
+ self._lease = bots_pb2.Lease()
|
|
| 163 |
+ self._lease.id = self._name
|
|
| 164 |
+ self._lease.payload.Pack(self.__operation_metadata.action_digest)
|
|
| 165 |
+ self._lease.state = LeaseState.PENDING.value
|
|
| 109 | 166 |
|
| 110 |
- def get_operation_meta(self):
|
|
| 111 |
- meta = remote_execution_pb2.ExecuteOperationMetadata()
|
|
| 112 |
- meta.stage = self._execute_stage.value
|
|
| 113 |
- meta.action_digest.CopyFrom(self._action_digest)
|
|
| 167 |
+ return self._lease
|
|
| 114 | 168 |
|
| 115 |
- return meta
|
|
| 169 |
+ def update_lease_state(self, state, status=None, result=None):
|
|
| 170 |
+ """Operates a state transition for the job's current :class:Lease.
|
|
| 116 | 171 |
|
| 117 |
- def create_lease(self):
|
|
| 118 |
- lease = bots_pb2.Lease(id=self.name, state=LeaseState.PENDING.value)
|
|
| 119 |
- lease.payload.Pack(self._action_digest)
|
|
| 172 |
+ Args:
|
|
| 173 |
+ state (LeaseState): the lease state to transition to.
|
|
| 174 |
+ status (google.rpc.Status): the lease execution status, only
|
|
| 175 |
+ required if `state` is `COMPLETED`.
|
|
| 176 |
+ result (google.protobuf.Any): the lease execution result, only
|
|
| 177 |
+ required if `state` is `COMPLETED`.
|
|
| 178 |
+ """
|
|
| 179 |
+ if state.value == self._lease.state:
|
|
| 180 |
+ return
|
|
| 181 |
+ |
|
| 182 |
+ self._lease.state = state.value
|
|
| 183 |
+ |
|
| 184 |
+ if self._lease.state == LeaseState.PENDING.value:
|
|
| 185 |
+ self.__worker_start_timestamp.Clear()
|
|
| 186 |
+ self.__worker_completed_timestamp.Clear()
|
|
| 187 |
+ |
|
| 188 |
+ self._lease.status.Clear()
|
|
| 189 |
+ self._lease.result.Clear()
|
|
| 120 | 190 |
|
| 121 |
- self.lease = lease
|
|
| 122 |
- return lease
|
|
| 191 |
+ elif self._lease.state == LeaseState.ACTIVE.value:
|
|
| 192 |
+ self.__worker_start_timestamp.GetCurrentTime()
|
|
| 123 | 193 |
|
| 124 |
- def get_operations(self):
|
|
| 125 |
- return operations_pb2.ListOperationsResponse(operations=[self.get_operation()])
|
|
| 194 |
+ elif self._lease.state == LeaseState.COMPLETED.value:
|
|
| 195 |
+ self.__worker_completed_timestamp.GetCurrentTime()
|
|
| 196 |
+ |
|
| 197 |
+ action_result = remote_execution_pb2.ActionResult()
|
|
| 198 |
+ |
|
| 199 |
+ # TODO: Make a distinction between build and bot failures!
|
|
| 200 |
+ if status.code != 0:
|
|
| 201 |
+ self._do_not_cache = True
|
|
| 202 |
+ |
|
| 203 |
+ if result is not None:
|
|
| 204 |
+ assert result.Is(action_result.DESCRIPTOR)
|
|
| 205 |
+ result.Unpack(action_result)
|
|
| 206 |
+ |
|
| 207 |
+ action_metadata = action_result.execution_metadata
|
|
| 208 |
+ action_metadata.queued_timestamp.CopyFrom(self.__worker_start_timestamp)
|
|
| 209 |
+ action_metadata.worker_start_timestamp.CopyFrom(self.__worker_start_timestamp)
|
|
| 210 |
+ action_metadata.worker_completed_timestamp.CopyFrom(self.__worker_completed_timestamp)
|
|
| 211 |
+ |
|
| 212 |
+ self.__execute_response = remote_execution_pb2.ExecuteResponse()
|
|
| 213 |
+ self.__execute_response.result.CopyFrom(action_result)
|
|
| 214 |
+ self.__execute_response.cached_result = False
|
|
| 215 |
+ self.__execute_response.status.CopyFrom(status)
|
|
| 126 | 216 |
|
| 127 | 217 |
def update_operation_stage(self, stage):
|
| 128 |
- self._execute_stage = stage
|
|
| 218 |
+ """Operates a stage transition for the job's :class:Operation.
|
|
| 219 |
+ |
|
| 220 |
+ Args:
|
|
| 221 |
+ stage (OperationStage): the operation stage to transition to.
|
|
| 222 |
+ """
|
|
| 223 |
+ if stage.value == self.__operation_metadata.stage:
|
|
| 224 |
+ return
|
|
| 225 |
+ |
|
| 226 |
+ self.__operation_metadata.stage = stage.value
|
|
| 227 |
+ |
|
| 228 |
+ if self.__operation_metadata.stage == OperationStage.QUEUED.value:
|
|
| 229 |
+ if self.__queued_timestamp.ByteSize() == 0:
|
|
| 230 |
+ self.__queued_timestamp.GetCurrentTime()
|
|
| 231 |
+ self._n_tries += 1
|
|
| 232 |
+ |
|
| 233 |
+ elif self.__operation_metadata.stage == OperationStage.COMPLETED.value:
|
|
| 234 |
+ if self.__execute_response is not None:
|
|
| 235 |
+ self._operation.response.Pack(self.__execute_response)
|
|
| 236 |
+ self._operation.done = True
|
|
| 237 |
+ |
|
| 238 |
+ self._operation.metadata.Pack(self.__operation_metadata)
|
|
| 239 |
+ |
|
| 129 | 240 |
for queue in self._operation_update_queues:
|
| 130 |
- queue.put(self.get_operation())
|
|
| 241 |
+ queue.put(self._operation)
|
| ... | ... | @@ -22,6 +22,7 @@ An instance of the LongRunningOperations Service. |
| 22 | 22 |
import logging
|
| 23 | 23 |
|
| 24 | 24 |
from buildgrid._exceptions import InvalidArgumentError
|
| 25 |
+from buildgrid._protos.google.longrunning import operations_pb2
|
|
| 25 | 26 |
|
| 26 | 27 |
|
| 27 | 28 |
class OperationsInstance:
|
| ... | ... | @@ -34,18 +35,21 @@ class OperationsInstance: |
| 34 | 35 |
server.add_operations_instance(self, instance_name)
|
| 35 | 36 |
|
| 36 | 37 |
def get_operation(self, name):
|
| 37 |
- operation = self._scheduler.jobs.get(name)
|
|
| 38 |
+ job = self._scheduler.jobs.get(name)
|
|
| 38 | 39 |
|
| 39 |
- if operation is None:
|
|
| 40 |
+ if job is None:
|
|
| 40 | 41 |
raise InvalidArgumentError("Operation name does not exist: [{}]".format(name))
|
| 41 | 42 |
|
| 42 | 43 |
else:
|
| 43 |
- return operation.get_operation()
|
|
| 44 |
+ return job.operation
|
|
| 44 | 45 |
|
| 45 | 46 |
def list_operations(self, list_filter, page_size, page_token):
|
| 46 | 47 |
# TODO: Pages
|
| 47 | 48 |
# Spec says number of pages and length of a page are optional
|
| 48 |
- return self._scheduler.get_operations()
|
|
| 49 |
+ response = operations_pb2.ListOperationsResponse()
|
|
| 50 |
+ response.operations.extend([job.operation for job in self._scheduler.list_jobs()])
|
|
| 51 |
+ |
|
| 52 |
+ return response
|
|
| 49 | 53 |
|
| 50 | 54 |
def delete_operation(self, name):
|
| 51 | 55 |
try:
|
| 1 |
-# Copyright (C) 2018 Codethink Limited
|
|
| 1 |
+# Copyright (C) 2018 Bloomberg LP
|
|
| 2 | 2 |
#
|
| 3 | 3 |
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 4 | 4 |
# you may not use this file except in compliance with the License.
|
| ... | ... | @@ -11,9 +11,7 @@ |
| 11 | 11 |
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 12 | 12 |
# See the License for the specific language governing permissions and
|
| 13 | 13 |
# limitations under the License.
|
| 14 |
-#
|
|
| 15 |
-# Authors:
|
|
| 16 |
-# Finn Ball <finn ball codethink co uk>
|
|
| 14 |
+ |
|
| 17 | 15 |
|
| 18 | 16 |
"""
|
| 19 | 17 |
Scheduler
|
| ... | ... | @@ -24,8 +22,6 @@ Schedules jobs. |
| 24 | 22 |
from collections import deque
|
| 25 | 23 |
|
| 26 | 24 |
from buildgrid._exceptions import NotFoundError
|
| 27 |
-from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
|
|
| 28 |
-from buildgrid._protos.google.longrunning import operations_pb2
|
|
| 29 | 25 |
|
| 30 | 26 |
from .job import OperationStage, LeaseState
|
| 31 | 27 |
|
| ... | ... | @@ -39,80 +35,101 @@ class Scheduler: |
| 39 | 35 |
self.jobs = {}
|
| 40 | 36 |
self.queue = deque()
|
| 41 | 37 |
|
| 42 |
- def register_client(self, name, queue):
|
|
| 43 |
- self.jobs[name].register_client(queue)
|
|
| 38 |
+ def register_client(self, job_name, queue):
|
|
| 39 |
+ self.jobs[job_name].register_client(queue)
|
|
| 44 | 40 |
|
| 45 |
- def unregister_client(self, name, queue):
|
|
| 46 |
- job = self.jobs[name]
|
|
| 47 |
- job.unregister_client(queue)
|
|
| 48 |
- if job.check_job_finished():
|
|
| 49 |
- del self.jobs[name]
|
|
| 41 |
+ def unregister_client(self, job_name, queue):
|
|
| 42 |
+ self.jobs[job_name].unregister_client(queue)
|
|
| 50 | 43 |
|
| 51 |
- def append_job(self, job, skip_cache_lookup=False):
|
|
| 44 |
+ if not self.jobs[job_name].n_clients and self.jobs[job_name].operation.done:
|
|
| 45 |
+ del self.jobs[job_name]
|
|
| 46 |
+ |
|
| 47 |
+ def queue_job(self, job, skip_cache_lookup=False):
|
|
| 52 | 48 |
self.jobs[job.name] = job
|
| 49 |
+ |
|
| 50 |
+ operation_stage = None
|
|
| 53 | 51 |
if self._action_cache is not None and not skip_cache_lookup:
|
| 54 | 52 |
try:
|
| 55 |
- cached_result = self._action_cache.get_action_result(job.action_digest)
|
|
| 53 |
+ action_result = self._action_cache.get_action_result(job.action_digest)
|
|
| 56 | 54 |
except NotFoundError:
|
| 55 |
+ operation_stage = OperationStage.QUEUED
|
|
| 57 | 56 |
self.queue.append(job)
|
| 58 |
- job.update_operation_stage(OperationStage.QUEUED)
|
|
| 59 | 57 |
|
| 60 | 58 |
else:
|
| 61 |
- job.result = cached_result
|
|
| 62 |
- job.result_cached = True
|
|
| 63 |
- job.update_operation_stage(OperationStage.COMPLETED)
|
|
| 59 |
+ job.set_cached_result(action_result)
|
|
| 60 |
+ operation_stage = OperationStage.COMPLETED
|
|
| 64 | 61 |
|
| 65 | 62 |
else:
|
| 63 |
+ operation_stage = OperationStage.QUEUED
|
|
| 66 | 64 |
self.queue.append(job)
|
| 67 |
- job.update_operation_stage(OperationStage.QUEUED)
|
|
| 68 | 65 |
|
| 69 |
- def retry_job(self, name):
|
|
| 70 |
- if name in self.jobs:
|
|
| 71 |
- job = self.jobs[name]
|
|
| 66 |
+ job.update_operation_stage(operation_stage)
|
|
| 67 |
+ |
|
| 68 |
+ def retry_job(self, job_name):
|
|
| 69 |
+ if job_name in self.jobs:
|
|
| 70 |
+ job = self.jobs[job_name]
|
|
| 72 | 71 |
if job.n_tries >= self.MAX_N_TRIES:
|
| 73 | 72 |
# TODO: Decide what to do with these jobs
|
| 74 | 73 |
job.update_operation_stage(OperationStage.COMPLETED)
|
| 75 | 74 |
# TODO: Mark these jobs as done
|
| 76 | 75 |
else:
|
| 77 | 76 |
job.update_operation_stage(OperationStage.QUEUED)
|
| 78 |
- job.n_tries += 1
|
|
| 79 | 77 |
self.queue.appendleft(job)
|
| 80 | 78 |
|
| 81 |
- def job_complete(self, name, result, status):
|
|
| 82 |
- job = self.jobs[name]
|
|
| 83 |
- job.lease.status.CopyFrom(status)
|
|
| 84 |
- action_result = remote_execution_pb2.ActionResult()
|
|
| 85 |
- result.Unpack(action_result)
|
|
| 86 |
- job.result = action_result
|
|
| 87 |
- if not job.do_not_cache and self._action_cache is not None:
|
|
| 88 |
- if not job.lease.status.code:
|
|
| 89 |
- self._action_cache.update_action_result(job.action_digest, action_result)
|
|
| 90 |
- job.update_operation_stage(OperationStage.COMPLETED)
|
|
| 91 |
- |
|
| 92 |
- def get_operations(self):
|
|
| 93 |
- response = operations_pb2.ListOperationsResponse()
|
|
| 94 |
- for v in self.jobs.values():
|
|
| 95 |
- response.operations.extend([v.get_operation()])
|
|
| 96 |
- return response
|
|
| 97 |
- |
|
| 98 |
- def update_job_lease_state(self, name, state):
|
|
| 99 |
- job = self.jobs[name]
|
|
| 100 |
- job.lease.state = state
|
|
| 101 |
- |
|
| 102 |
- def get_job_lease(self, name):
|
|
| 103 |
- return self.jobs[name].lease
|
|
| 104 |
- |
|
| 105 |
- def cancel_session(self, name):
|
|
| 106 |
- job = self.jobs[name]
|
|
| 107 |
- state = job.lease.state
|
|
| 108 |
- if state in (LeaseState.PENDING.value, LeaseState.ACTIVE.value):
|
|
| 109 |
- self.retry_job(name)
|
|
| 110 |
- |
|
| 111 |
- def create_lease(self):
|
|
| 112 |
- if self.queue:
|
|
| 113 |
- job = self.queue.popleft()
|
|
| 79 |
+ def list_jobs(self):
|
|
| 80 |
+ return self.jobs.values()
|
|
| 81 |
+ |
|
| 82 |
+ def request_job_leases(self, worker_capabilities):
|
|
| 83 |
+ """Generates a list of the highest priority leases to be run.
|
|
| 84 |
+ |
|
| 85 |
+ Args:
|
|
| 86 |
+ worker_capabilities (dict): a set of key-value pairs decribing the
|
|
| 87 |
+ worker properties, configuration and state at the time of the
|
|
| 88 |
+ request.
|
|
| 89 |
+ """
|
|
| 90 |
+ if not self.queue:
|
|
| 91 |
+ return []
|
|
| 92 |
+ |
|
| 93 |
+ job = self.queue.popleft()
|
|
| 94 |
+ # For now, one lease at a time:
|
|
| 95 |
+ lease = job.create_lease()
|
|
| 96 |
+ |
|
| 97 |
+ return [lease]
|
|
| 98 |
+ |
|
| 99 |
+ def update_job_lease_state(self, job_name, lease_state, lease_status=None, lease_result=None):
|
|
| 100 |
+ """Requests a state transition for a job's current :class:Lease.
|
|
| 101 |
+ |
|
| 102 |
+ Args:
|
|
| 103 |
+ job_name (str): name of the job to query.
|
|
| 104 |
+ lease_state (LeaseState): the lease state to transition to.
|
|
| 105 |
+ lease_status (google.rpc.Status): the lease execution status, only
|
|
| 106 |
+ required if `lease_state` is `COMPLETED`.
|
|
| 107 |
+ lease_result (google.protobuf.Any): the lease execution result, only
|
|
| 108 |
+ required if `lease_state` is `COMPLETED`.
|
|
| 109 |
+ """
|
|
| 110 |
+ job = self.jobs[job_name]
|
|
| 111 |
+ |
|
| 112 |
+ if lease_state == LeaseState.PENDING:
|
|
| 113 |
+ job.update_lease_state(LeaseState.PENDING)
|
|
| 114 |
+ job.update_operation_stage(OperationStage.QUEUED)
|
|
| 115 |
+ |
|
| 116 |
+ elif lease_state == LeaseState.ACTIVE:
|
|
| 117 |
+ job.update_lease_state(LeaseState.ACTIVE)
|
|
| 114 | 118 |
job.update_operation_stage(OperationStage.EXECUTING)
|
| 115 |
- job.create_lease()
|
|
| 116 |
- job.lease.state = LeaseState.PENDING.value
|
|
| 117 |
- return job.lease
|
|
| 118 |
- return None
|
|
| 119 |
+ |
|
| 120 |
+ elif lease_state == LeaseState.COMPLETED:
|
|
| 121 |
+ job.update_lease_state(LeaseState.COMPLETED,
|
|
| 122 |
+ status=lease_status, result=lease_result)
|
|
| 123 |
+ |
|
| 124 |
+ if self._action_cache is not None and not job.do_not_cache:
|
|
| 125 |
+ self._action_cache.update_action_result(job.action_digest, job.action_result)
|
|
| 126 |
+ |
|
| 127 |
+ job.update_operation_stage(OperationStage.COMPLETED)
|
|
| 128 |
+ |
|
| 129 |
+ def get_job_lease(self, job_name):
|
|
| 130 |
+ """Returns the lease associated to job, if any have been emitted yet."""
|
|
| 131 |
+ return self.jobs[job_name].lease
|
|
| 132 |
+ |
|
| 133 |
+ def get_job_operation(self, job_name):
|
|
| 134 |
+ """Returns the operation associated to job."""
|
|
| 135 |
+ return self.jobs[job_name].operation
|
| ... | ... | @@ -15,11 +15,21 @@ |
| 15 | 15 |
|
| 16 | 16 |
from operator import attrgetter
|
| 17 | 17 |
import os
|
| 18 |
+import socket
|
|
| 18 | 19 |
|
| 19 | 20 |
from buildgrid.settings import HASH
|
| 20 | 21 |
from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
|
| 21 | 22 |
|
| 22 | 23 |
|
| 24 |
+def get_hostname():
|
|
| 25 |
+ """Returns the hostname of the machine executing that function.
|
|
| 26 |
+ |
|
| 27 |
+ Returns:
|
|
| 28 |
+ str: Hostname for the current machine.
|
|
| 29 |
+ """
|
|
| 30 |
+ return socket.gethostname()
|
|
| 31 |
+ |
|
| 32 |
+ |
|
| 23 | 33 |
def create_digest(bytes_to_digest):
|
| 24 | 34 |
"""Computes the :obj:`Digest` of a piece of data.
|
| 25 | 35 |
|
| ... | ... | @@ -119,7 +119,6 @@ setup( |
| 119 | 119 |
'pyaml',
|
| 120 | 120 |
'boto3 < 1.8.0',
|
| 121 | 121 |
'botocore < 1.11.0',
|
| 122 |
- 'xdg',
|
|
| 123 | 122 |
],
|
| 124 | 123 |
entry_points={
|
| 125 | 124 |
'console_scripts': [
|
| ... | ... | @@ -137,7 +137,7 @@ def test_update_leases_with_work(bot_session, context, instance): |
| 137 | 137 |
bot_session=bot_session)
|
| 138 | 138 |
|
| 139 | 139 |
action_digest = remote_execution_pb2.Digest(hash='gaff')
|
| 140 |
- _inject_work(instance._instances[""]._scheduler, action_digest)
|
|
| 140 |
+ _inject_work(instance._instances[""]._scheduler, action_digest=action_digest)
|
|
| 141 | 141 |
|
| 142 | 142 |
response = instance.CreateBotSession(request, context)
|
| 143 | 143 |
|
| ... | ... | @@ -159,7 +159,7 @@ def test_update_leases_work_complete(bot_session, context, instance): |
| 159 | 159 |
|
| 160 | 160 |
# Inject work
|
| 161 | 161 |
action_digest = remote_execution_pb2.Digest(hash='gaff')
|
| 162 |
- _inject_work(instance._instances[""]._scheduler, action_digest)
|
|
| 162 |
+ _inject_work(instance._instances[""]._scheduler, action_digest=action_digest)
|
|
| 163 | 163 |
|
| 164 | 164 |
request = bots_pb2.UpdateBotSessionRequest(name=response.name,
|
| 165 | 165 |
bot_session=response)
|
| ... | ... | @@ -174,6 +174,7 @@ def test_update_leases_work_complete(bot_session, context, instance): |
| 174 | 174 |
response = copy.deepcopy(instance.UpdateBotSession(request, context))
|
| 175 | 175 |
|
| 176 | 176 |
response.leases[0].state = LeaseState.COMPLETED.value
|
| 177 |
+ response.leases[0].result.Pack(remote_execution_pb2.ActionResult())
|
|
| 177 | 178 |
|
| 178 | 179 |
request = bots_pb2.UpdateBotSessionRequest(name=response.name,
|
| 179 | 180 |
bot_session=response)
|
| ... | ... | @@ -187,7 +188,7 @@ def test_work_rejected_by_bot(bot_session, context, instance): |
| 187 | 188 |
bot_session=bot_session)
|
| 188 | 189 |
# Inject work
|
| 189 | 190 |
action_digest = remote_execution_pb2.Digest(hash='gaff')
|
| 190 |
- _inject_work(instance._instances[""]._scheduler, action_digest)
|
|
| 191 |
+ _inject_work(instance._instances[""]._scheduler, action_digest=action_digest)
|
|
| 191 | 192 |
|
| 192 | 193 |
# Simulated the severed binding between client and server
|
| 193 | 194 |
response = copy.deepcopy(instance.CreateBotSession(request, context))
|
| ... | ... | @@ -209,7 +210,7 @@ def test_work_out_of_sync_from_pending(state, bot_session, context, instance): |
| 209 | 210 |
bot_session=bot_session)
|
| 210 | 211 |
# Inject work
|
| 211 | 212 |
action_digest = remote_execution_pb2.Digest(hash='gaff')
|
| 212 |
- _inject_work(instance._instances[""]._scheduler, action_digest)
|
|
| 213 |
+ _inject_work(instance._instances[""]._scheduler, action_digest=action_digest)
|
|
| 213 | 214 |
|
| 214 | 215 |
# Simulated the severed binding between client and server
|
| 215 | 216 |
response = copy.deepcopy(instance.CreateBotSession(request, context))
|
| ... | ... | @@ -230,7 +231,7 @@ def test_work_out_of_sync_from_active(state, bot_session, context, instance): |
| 230 | 231 |
bot_session=bot_session)
|
| 231 | 232 |
# Inject work
|
| 232 | 233 |
action_digest = remote_execution_pb2.Digest(hash='gaff')
|
| 233 |
- _inject_work(instance._instances[""]._scheduler, action_digest)
|
|
| 234 |
+ _inject_work(instance._instances[""]._scheduler, action_digest=action_digest)
|
|
| 234 | 235 |
|
| 235 | 236 |
# Simulated the severed binding between client and server
|
| 236 | 237 |
response = copy.deepcopy(instance.CreateBotSession(request, context))
|
| ... | ... | @@ -257,7 +258,7 @@ def test_work_active_to_active(bot_session, context, instance): |
| 257 | 258 |
bot_session=bot_session)
|
| 258 | 259 |
# Inject work
|
| 259 | 260 |
action_digest = remote_execution_pb2.Digest(hash='gaff')
|
| 260 |
- _inject_work(instance._instances[""]._scheduler, action_digest)
|
|
| 261 |
+ _inject_work(instance._instances[""]._scheduler, action_digest=action_digest)
|
|
| 261 | 262 |
|
| 262 | 263 |
# Simulated the severed binding between client and server
|
| 263 | 264 |
response = copy.deepcopy(instance.CreateBotSession(request, context))
|
| ... | ... | @@ -279,8 +280,10 @@ def test_post_bot_event_temp(context, instance): |
| 279 | 280 |
context.set_code.assert_called_once_with(grpc.StatusCode.UNIMPLEMENTED)
|
| 280 | 281 |
|
| 281 | 282 |
|
| 282 |
-def _inject_work(scheduler, action_digest=None):
|
|
| 283 |
+def _inject_work(scheduler, action=None, action_digest=None):
|
|
| 284 |
+ if not action:
|
|
| 285 |
+ action = remote_execution_pb2.Action()
|
|
| 283 | 286 |
if not action_digest:
|
| 284 | 287 |
action_digest = remote_execution_pb2.Digest()
|
| 285 |
- j = job.Job(action_digest, False)
|
|
| 286 |
- scheduler.append_job(j, True)
|
|
| 288 |
+ j = job.Job(action, action_digest)
|
|
| 289 |
+ scheduler.queue_job(j, True)
|
| ... | ... | @@ -105,7 +105,7 @@ def test_no_action_digest_in_storage(instance, context): |
| 105 | 105 |
|
| 106 | 106 |
|
| 107 | 107 |
def test_wait_execution(instance, controller, context):
|
| 108 |
- j = job.Job(action_digest, None)
|
|
| 108 |
+ j = job.Job(action, action_digest)
|
|
| 109 | 109 |
j._operation.done = True
|
| 110 | 110 |
|
| 111 | 111 |
request = remote_execution_pb2.WaitExecutionRequest(name="{}/{}".format('', j.name))
|
| ... | ... | @@ -30,6 +30,7 @@ from buildgrid._protos.google.longrunning import operations_pb2 |
| 30 | 30 |
from buildgrid._protos.google.rpc import status_pb2
|
| 31 | 31 |
from buildgrid.server.cas.storage import lru_memory_cache
|
| 32 | 32 |
from buildgrid.server.controller import ExecutionController
|
| 33 |
+from buildgrid.server.job import LeaseState
|
|
| 33 | 34 |
from buildgrid.server.operations import service
|
| 34 | 35 |
from buildgrid.server.operations.service import OperationsService
|
| 35 | 36 |
from buildgrid.utils import create_digest
|
| ... | ... | @@ -131,9 +132,10 @@ def test_list_operations_with_result(instance, controller, execute_request, cont |
| 131 | 132 |
action_result.output_files.extend([output_file])
|
| 132 | 133 |
|
| 133 | 134 |
controller.operations_instance._scheduler.jobs[response_execute.name].create_lease()
|
| 134 |
- controller.operations_instance._scheduler.job_complete(response_execute.name,
|
|
| 135 |
- _pack_any(action_result),
|
|
| 136 |
- status_pb2.Status())
|
|
| 135 |
+ controller.operations_instance._scheduler.update_job_lease_state(response_execute.name,
|
|
| 136 |
+ LeaseState.COMPLETED,
|
|
| 137 |
+ lease_status=status_pb2.Status(),
|
|
| 138 |
+ lease_result=_pack_any(action_result))
|
|
| 137 | 139 |
|
| 138 | 140 |
request = operations_pb2.ListOperationsRequest(name=instance_name)
|
| 139 | 141 |
response = instance.ListOperations(request, context)
|
| ... | ... | @@ -142,7 +144,8 @@ def test_list_operations_with_result(instance, controller, execute_request, cont |
| 142 | 144 |
|
| 143 | 145 |
execute_response = remote_execution_pb2.ExecuteResponse()
|
| 144 | 146 |
response.operations[0].response.Unpack(execute_response)
|
| 145 |
- assert execute_response.result == action_result
|
|
| 147 |
+ |
|
| 148 |
+ assert execute_response.result.output_files == action_result.output_files
|
|
| 146 | 149 |
|
| 147 | 150 |
|
| 148 | 151 |
def test_list_operations_empty(instance, context):
|
