GNOME.org
 

[Notes] [Git][BuildGrid/buildgrid][mablanch/83-executed-action-metadata] 24 commits: setup.py: Unpin pytest version but require >= 3.8.0

Title: GitLab

Martin Blanchard pushed to branch mablanch/83-executed-action-metadata at BuildGrid / buildgrid

Commits:

22 changed files:

Changes:

  • .gitlab-ci.yml
    ... ... @@ -43,7 +43,6 @@ tests-debian-stretch:
    43 43 
       <<: *linux-tests
    44 44
    45 45 
     run-dummy-job-debian:
    46 
    -  image: buildstream/buildstream-debian
    47 46 
       <<: *dummy-job
    48 47
    49 48

  • .pylintrc
    ... ... @@ -455,8 +455,12 @@ known-standard-library=
    455 455
    456 456 
     # Force import order to recognize a module as part of a third party library.
    457 457 
     known-third-party=boto3,
    458 
    +                  click,
    458 459 
                       enchant,
    459 
    -                  grpc
    460 
    +                  google,
    461 
    +                  grpc,
    462 
    +                  moto,
    463 
    +                  yaml
    460 464
    461 465
    462 466 
     [DESIGN]

  • CONTRIBUTING.rst
    1 
    -
    2 1 
     .. _contributing:
    3 2
    4 3 
     Contributing
    ... ... @@ -18,9 +17,13 @@ discuss with us before submitting anything, as we may well have some important
    18 17 
     context which will could help guide your efforts.
    19 18
    20 19 
     Any major feature additions should be raised first as a proposal on the
    21 
    -`BuildGrid mailing list`_ to be discussed, and then eventually followed up with
    22 
    -an issue on GitLab. We recommend that you propose the feature in advance of
    23 
    -commencing work.
    20 
    +BuildGrid mailing list to be discussed between the core contributors. Once
    21 
    +this discussion has taken place and there is agreement on how to proceed,
    22 
    +it should be followed by with a Gitlab issue being raised which summarizes
    23 
    +the tasks required.
    24 
    +
    25 
    +We strongly recommend that you propose the feature in advance of
    26 
    +commencing any work.
    24 27
    25 28 
     The author of any patch is expected to take ownership of that code and is to
    26 29 
     support it for a reasonable time-frame. This means addressing any unforeseen
    ... ... @@ -204,17 +207,16 @@ Committer access
    204 207 
     We'll hand out commit access to anyone who has successfully landed a single
    205 208 
     patch to the code base. Please request this via Slack or the mailing list.
    206 209
    207 
    -This of course relies on contributors being responsive and show willingness to
    208 
    -address problems after landing branches there should not be any problems here.
    210 
    +This of course relies on contributors being responsive and showing willingness
    211 
    +to address any problems that may arise after landing branches.
    209 212
    210 
    -What we are expecting of committers here in general is basically to escalate the
    211 
    -review in cases of uncertainty:
    213 
    +When submitting a merge request, please obtain a review from another committer
    214 
    +who is familiar with the area of the code base which the branch effects. An
    215 
    +approval from another committer who is not the patch author will be needed
    216 
    +before any merge (we use gitlab's 'approval' feature for this).
    212 217
    213 
    -- If the change is very trivial (obvious few line changes, typos…), and you are
    214 
    -  confident of the change, there is no need for review.
    215 
    -- If the change is non trivial, please obtain a review from another committer
    216 
    -  who is familiar with the area which the branch effects. An approval from
    217 
    -  someone who is not the patch author will be needed before any merge.
    218 
    +What we are expecting of committers here in general is basically to escalate the
    219 
    +review in cases of uncertainty.
    218 220
    219 221 
     .. note::
    220 222
    ... ... @@ -239,21 +241,49 @@ following goals:
    239 241 
       for the viewer to digest.
    240 242 
     - Ensure that we keep it simple and easy to contribute to the project.
    241 243
    242 
    -We are currenlty using the following GitLab features:
    244 
    +Explanation of how the project is currenlty using some GitLab features:
    243 245
    244 246 
     - `Milestones`_: we have seen them used in the same way as `Epics`_ in other
    245 
    -  projects. BuildGrid milestones must be time-line based, can overlap and we can
    246 
    -  be working towards multiple milestones at any one time. They allow us to group
    247 
    -  together all sub tasks into an overall aim. See our `BuildGrid milestones`_.
    248 
    -- `Labels`_: allow us to filter tickets in useful ways. They do complexity and
    249 
    -  effort as they grow in number and usage, though, so the general approach is
    250 
    -  to have the minimum possible. See our `BuildGrid labels`_.
    247 
    +  projects and are trying not to do that here. Instead we are going to
    248 
    +  use milestones to denote development cycles (ie, two week 'sprints'). See the
    249 
    +  `BuildGrid milestones`_.
    250 
    +- `Labels`_: allow us to filter tickets (ie, 'issues' in gitlab terminology)
    251 
    +  in useful ways. They add complexity and effort as they grow in number, so the
    252 
    +  general approach is to have the minimum possible but
    253 
    +  ensure we use them consistently. See the `BuildGrid labels`_.
    251 254 
     - `Boards`_: allow us to visualise and manage issues and labels in a simple way.
    252 
    -  For now, we are only utilising one boards. Issues start life in the
    253 
    -  ``Backlog`` column by default, and we move them into ``ToDo`` when they are
    254 
    -  coming up in the next few weeks. ``Doing`` is only for when an item is
    255 
    -  currently being worked on. Moving an issue from column to column automatically
    256 
    -  adjust the tagged labels. See our `BuildGrid boards`_.
    255 
    +  Issues start life in the ``Backlog`` column by default, and we move them into
    256 
    +  ``ToDo`` when we aim to complete them in the current development cycle.
    257 
    +  ``Doing`` is only for when an item is currently being worked on. When on the
    258 
    +  Board view, dragging and dropping an issue from column to column automatically
    259 
    +  adjusts the relevant labels. See the `BuildGrid boards`_.
    260 
    +
    261 
    +
    262 
    +Guidelines for using GitLab features when working on this project:
    263 
    +
    264 
    +- When raising an issue, please:
    265 
    +
    266 
    +  - check to see if there already is an issue to cover this task (if not then
    267 
    +    raise a new one)
    268 
    +  - assign the appropriate label or labels (tip: the vast majority of issues
    269 
    +    raised will be either an enhancement or a bug)
    270 
    +
    271 
    +- If you plan to work on an issue, please:
    272 
    +
    273 
    +  - self-assign the ticket
    274 
    +  - ensure it's captured in the current sprint (ie, Gitlab milestone)
    275 
    +  - ensure the ticket is in the ``ToDo`` column of the board if you aim to
    276 
    +    complete in the current sprint but aren't yet working on it, and
    277 
    +    the ``Doing`` column if you are working on it currently.
    278 
    +
    279 
    +- Please note that Gitlab issues are for either 'tasks' or 'bugs' - ie not for
    280 
    +  long discussions (where the mailing list is a better choice) or for ranting,
    281 
    +  for example.
    282 
    +
    283 
    +The above may seem like a lot to take in, but please don't worry about getting
    284 
    +it right the first few times. The worst that can happen is that you'll get a
    285 
    +friendly message from a current contributor who explains the process. We welcome
    286 
    +and value all contributions to the project!
    257 287
    258 288 
     .. _Milestones: https://docs.gitlab.com/ee/user/project/milestones
    259 289 
     .. _Epics: https://docs.gitlab.com/ee/user/group/epics

  • Dockerfile
    1 
    +FROM python:3.5-stretch
    2 
    +
    3 
    +# Point the path to where buildgrid gets installed
    4 
    +ENV PATH=$PATH:/root/.local/bin/
    5 
    +
    6 
    +# Upgrade python modules
    7 
    +RUN python3 -m pip install --upgrade setuptools pip
    8 
    +
    9 
    +# Use /app as the current working directory
    10 
    +WORKDIR /app
    11 
    +
    12 
    +# Copy the repo contents (source, config files, etc) in the WORKDIR
    13 
    +COPY . .
    14 
    +
    15 
    +# Install BuildGrid
    16 
    +RUN pip install --user --editable .
    17 
    +
    18 
    +# Entry Point of the image (should get an additional argument from CMD, the path to the config file)
    19 
    +ENTRYPOINT ["bgd", "-v", "server", "start"]
    20 
    +
    21 
    +# Default config file (used if no CMD specified when running)
    22 
    +CMD ["buildgrid/_app/settings/default.yml"]
    23 
    +

  • README.rst
    1 
    -
    2 1 
     .. _about:
    3 2
    4 3 
     About
    ... ... @@ -14,13 +13,15 @@ BuildGrid is a Python remote execution service which implements Google's
    14 13 
     `Remote Execution API`_ and the `Remote Workers API`_. The project's goal is to
    15 14 
     be able to execute build jobs remotely on a grid of computers in order to
    16 15 
     massively speed up build times. Workers on the grid should be able to run with
    17 
    -different environments. It is designed to work with clients such as `Bazel`_ and
    18 
    -`BuildStream`_.
    16 
    +different environments. It works with clients such as `Bazel`_,
    17 
    +`BuildStream`_ and `RECC`_, and is designed to be able to work with any client
    18 
    +that conforms to the above API protocols.
    19 19
    20 20 
     .. _Remote Execution API: https://github.com/bazelbuild/remote-apis
    21 21 
     .. _Remote Workers API: https://docs.google.com/document/d/1s_AzRRD2mdyktKUj2HWBn99rMg_3tcPvdjx3MPbFidU/edit#heading=h.1u2taqr2h940
    22 22 
     .. _BuildStream: https://wiki.gnome.org/Projects/BuildStream
    23 23 
     .. _Bazel: https://bazel.build
    24 
    +.. _RECC: https://gitlab.com/bloomberg/recc
    24 25
    25 26
    26 27 
     .. _getting-started:
    ... ... @@ -49,7 +50,7 @@ Resources
    49 50
    50 51 
     .. _Homepage: https://buildgrid.build
    51 52 
     .. _GitLab repository: https://gitlab.com/BuildGrid/buildgrid
    52 
    -.. _Bug tracking: https://gitlab.com/BuildGrid/buildgrid/issues
    53 
    +.. _Bug tracking: https://gitlab.com/BuildGrid/buildgrid/boards
    53 54 
     .. _Mailing list: https://lists.buildgrid.build/cgi-bin/mailman/listinfo/buildgrid
    54 55 
     .. _Slack channel: https://buildteamworld.slack.com/messages/CC9MKC203
    55 56 
     .. _invite link: https://join.slack.com/t/buildteamworld/shared_invite/enQtMzkxNzE0MDMyMDY1LTRmZmM1OWE0OTFkMGE1YjU5Njc4ODEzYjc0MGMyOTM5ZTQ5MmE2YTQ1MzQwZDc5MWNhODY1ZmRkZTE4YjFhNjU

  • buildgrid/_app/bots/buildbox.py
    ... ... @@ -17,8 +17,6 @@ import os
    17 17 
     import subprocess
    18 18 
     import tempfile
    19 19
    20 
    -from google.protobuf import any_pb2
    21 
    -
    22 20 
     from buildgrid.client.cas import download, upload
    23 21 
     from buildgrid._exceptions import BotError
    24 22 
     from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
    ... ... @@ -29,13 +27,14 @@ from buildgrid.utils import read_file, write_file
    29 27 
     def work_buildbox(context, lease):
    30 28 
         """Executes a lease for a build action, using buildbox.
    31 29 
         """
    32 
    -
    33 30 
         local_cas_directory = context.local_cas
    34 31 
         # instance_name = context.parent
    35 32 
         logger = context.logger
    36 33
    37 34 
         action_digest = remote_execution_pb2.Digest()
    35 
    +
    38 36 
         lease.payload.Unpack(action_digest)
    37 
    +    lease.result.Clear()
    39 38
    40 39 
         with download(context.cas_channel) as downloader:
    41 40 
             action = downloader.get_message(action_digest,
    ... ... @@ -131,10 +130,7 @@ def work_buildbox(context, lease):
    131 130
    132 131 
                 action_result.output_directories.extend([output_directory])
    133 132
    134 
    -            action_result_any = any_pb2.Any()
    135 
    -            action_result_any.Pack(action_result)
    136 
    -
    137 
    -            lease.result.CopyFrom(action_result_any)
    133 
    +            lease.result.Pack(action_result)
    138 134
    139 135 
         return lease
    140 136

  • buildgrid/_app/bots/dummy.py
    ... ... @@ -16,9 +16,18 @@
    16 16 
     import random
    17 17 
     import time
    18 18
    19 
    +from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
    20 
    +
    19 21
    20 22 
     def work_dummy(context, lease):
    21 23 
         """ Just returns lease after some random time
    22 24 
         """
    25 
    +    lease.result.Clear()
    26 
    +
    23 27 
         time.sleep(random.randint(1, 5))
    28 
    +
    29 
    +    action_result = remote_execution_pb2.ActionResult()
    30 
    +
    31 
    +    lease.result.Pack(action_result)
    32 
    +
    24 33 
         return lease

  • buildgrid/_app/bots/host.py
    ... ... @@ -17,8 +17,6 @@ import os
    17 17 
     import subprocess
    18 18 
     import tempfile
    19 19
    20 
    -from google.protobuf import any_pb2
    21 
    -
    22 20 
     from buildgrid.client.cas import download, upload
    23 21 
     from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
    24 22 
     from buildgrid.utils import output_file_maker, output_directory_maker
    ... ... @@ -27,12 +25,13 @@ from buildgrid.utils import output_file_maker, output_directory_maker
    27 25 
     def work_host_tools(context, lease):
    28 26 
         """Executes a lease for a build action, using host tools.
    29 27 
         """
    30 
    -
    31 28 
         instance_name = context.parent
    32 29 
         logger = context.logger
    33 30
    34 31 
         action_digest = remote_execution_pb2.Digest()
    32 
    +
    35 33 
         lease.payload.Unpack(action_digest)
    34 
    +    lease.result.Clear()
    36 35
    37 36 
         with tempfile.TemporaryDirectory() as temp_directory:
    38 37 
             with download(context.cas_channel, instance=instance_name) as downloader:
    ... ... @@ -122,9 +121,6 @@ def work_host_tools(context, lease):
    122 121
    123 122 
                 action_result.output_directories.extend(output_directories)
    124 123
    125 
    -        action_result_any = any_pb2.Any()
    126 
    -        action_result_any.Pack(action_result)
    127 
    -
    128 
    -        lease.result.CopyFrom(action_result_any)
    124 
    +        lease.result.Pack(action_result)
    129 125
    130 126 
         return lease

  • buildgrid/_app/commands/cmd_execute.py
    ... ... @@ -169,6 +169,7 @@ def run_command(context, input_root, commands, output_file, output_directory):
    169 169
    170 170 
                 downloader.download_file(output_file_response.digest, path)
    171 171
    172 
    -            if output_file_response.path in output_executeables:
    173 
    -                st = os.stat(path)
    174 
    -                os.chmod(path, st.st_mode | stat.S_IXUSR)
    172 
    +    for output_file_response in execute_response.result.output_files:
    173 
    +        if output_file_response.path in output_executeables:
    174 
    +            st = os.stat(path)
    175 
    +            os.chmod(path, st.st_mode | stat.S_IXUSR)

  • buildgrid/server/bots/instance.py
    ... ... @@ -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:

  • buildgrid/server/execution/instance.py
    ... ... @@ -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:

  • buildgrid/server/job.py
    ... ... @@ -11,151 +11,230 @@
    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
    22 
    -from google.protobuf import any_pb2
    20 
    +from from google.protobuf import timestamp_pb2
    23 21
    24 22 
     from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
    25 23 
     from buildgrid._protos.google.devtools.remoteworkers.v1test2 import bots_pb2
    26 24 
     from buildgrid._protos.google.longrunning import operations_pb2
    27 25
    28 26
    29 
    -class ExecuteStage(Enum):
    27 
    +class OperationStage(Enum):
    28 
    +    # Initially unknown stage.
    30 29 
         UNKNOWN = remote_execution_pb2.ExecuteOperationMetadata.Stage.Value('UNKNOWN')
    31 
    -
    32 30 
         # Checking the result against the cache.
    33 31 
         CACHE_CHECK = remote_execution_pb2.ExecuteOperationMetadata.Stage.Value('CACHE_CHECK')
    34 
    -
    35 32 
         # Currently idle, awaiting a free machine to execute.
    36 33 
         QUEUED = remote_execution_pb2.ExecuteOperationMetadata.Stage.Value('QUEUED')
    37 
    -
    38 34 
         # Currently being executed by a worker.
    39 35 
         EXECUTING = remote_execution_pb2.ExecuteOperationMetadata.Stage.Value('EXECUTING')
    40 
    -
    41 36 
         # Finished execution.
    42 37 
         COMPLETED = remote_execution_pb2.ExecuteOperationMetadata.Stage.Value('COMPLETED')
    43 38
    44 39
    45 
    -class BotStatus(Enum):
    46 
    -    BOT_STATUS_UNSPECIFIED = bots_pb2.BotStatus.Value('BOT_STATUS_UNSPECIFIED')
    47 
    -
    48 
    -    # The bot is healthy, and will accept leases as normal.
    49 
    -    OK = bots_pb2.BotStatus.Value('OK')
    50 
    -
    51 
    -    # The bot is unhealthy and will not accept new leases.
    52 
    -    UNHEALTHY = bots_pb2.BotStatus.Value('UNHEALTHY')
    53 
    -
    54 
    -    # The bot has been asked to reboot the host.
    55 
    -    HOST_REBOOTING = bots_pb2.BotStatus.Value('HOST_REBOOTING')
    56 
    -
    57 
    -    # The bot has been asked to shut down.
    58 
    -    BOT_TERMINATING = bots_pb2.BotStatus.Value('BOT_TERMINATING')
    59 
    -
    60 
    -
    61 40 
     class LeaseState(Enum):
    41 
    +    # Initially unknown state.
    62 42 
         LEASE_STATE_UNSPECIFIED = bots_pb2.LeaseState.Value('LEASE_STATE_UNSPECIFIED')
    63 
    -
    64 43 
         # The server expects the bot to accept this lease.
    65 44 
         PENDING = bots_pb2.LeaseState.Value('PENDING')
    66 
    -
    67 45 
         # The bot has accepted this lease.
    68 46 
         ACTIVE = bots_pb2.LeaseState.Value('ACTIVE')
    69 
    -
    70 47 
         # The bot is no longer leased.
    71 48 
         COMPLETED = bots_pb2.LeaseState.Value('COMPLETED')
    72 
    -
    73 49 
         # The bot should immediately release all resources associated with the lease.
    74 50 
         CANCELLED = bots_pb2.LeaseState.Value('CANCELLED')
    75 51
    76 52
    77 53 
     class Job:
    78 54
    79 
    -    def __init__(self, action_digest, do_not_cache=False, message_queue=None):
    80 
    -        self.lease = None
    55 
    +    def __init__(self, action, action_digest):
    81 56 
             self.logger = logging.getLogger(__name__)
    82 
    -        self.n_tries = 0
    83 
    -        self.result = None
    84 
    -        self.result_cached = False
    85 57
    86 
    -        self._action_digest = action_digest
    87 
    -        self._do_not_cache = do_not_cache
    88 
    -        self._execute_stage = ExecuteStage.UNKNOWN
    89 58 
             self._name = str(uuid.uuid4())
    90 
    -        self._operation = operations_pb2.Operation(name=self._name)
    91 
    -        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()
    92 68
    93 
    -        if message_queue is not None:
    94 
    -            self.register_client(message_queue)
    69 
    +        self.__operation_metadata.action_digest.CopyFrom(action_digest)
    70 
    +        self.__operation_metadata.stage = OperationStage.UNKNOWN.value
    71 
    +
    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
    95 78
    96 79 
         @property
    97 80 
         def name(self):
    98 81 
             return self._name
    99 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 
    +
    100 91 
         @property
    101 92 
         def action_digest(self):
    102 
    -        return self._action_digest
    93 
    +        return self.__operation_metadata.action_digest
    103 94
    104 95 
         @property
    105 
    -    def do_not_cache(self):
    106 
    -        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
    107 105
    108 
    -    def check_job_finished(self):
    109 
    -        if not self._operation_update_queues:
    110 
    -            return self._operation.done
    111 
    -        return False
    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
    120 
    +
    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)
    112 128
    113 129 
         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 
    +        """
    114 135 
             self._operation_update_queues.append(queue)
    115 
    -        queue.put(self.get_operation())
    136 
    +        queue.put(self._operation)
    116 137
    117 138 
         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 
    +        """
    118 144 
             self._operation_update_queues.remove(queue)
    119 145
    120 
    -    def get_operation(self):
    121 
    -        self._operation.metadata.CopyFrom(self._pack_any(self.get_operation_meta()))
    122 
    -        if self.result is not None:
    123 
    -            self._operation.done = True
    124 
    -            response = remote_execution_pb2.ExecuteResponse(result=self.result,
    125 
    -                                                            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
    126 152
    127 
    -            if not self.result_cached:
    128 
    -                response.status.CopyFrom(self.lease.status)
    153 
    +    def create_lease(self):
    154 
    +        """Emits a new :class:`Lease` for the job.
    129 155
    130 
    -            self._operation.response.CopyFrom(self._pack_any(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
    131 161
    132 
    -        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
    133 166
    134 
    -    def get_operation_meta(self):
    135 
    -        meta = remote_execution_pb2.ExecuteOperationMetadata()
    136 
    -        meta.stage = self._execute_stage.value
    137 
    -        meta.action_digest.CopyFrom(self._action_digest)
    167 
    +        return self._lease
    138 168
    139 
    -        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.
    140 171
    141 
    -    def create_lease(self):
    142 
    -        action_digest = self._pack_any(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
    143 181
    144 
    -        lease = bots_pb2.Lease(id=self.name,
    145 
    -                               payload=action_digest,
    146 
    -                               state=LeaseState.PENDING.value)
    147 
    -        self.lease = lease
    148 
    -        return lease
    182 
    +        self._lease.state = state.value
    149 183
    150 
    -    def get_operations(self):
    151 
    -        return operations_pb2.ListOperationsResponse(operations=[self.get_operation()])
    184 
    +        if self._lease.state == LeaseState.PENDING.value:
    185 
    +            self.__worker_start_timestamp.Clear()
    186 
    +            self.__worker_completed_timestamp.Clear()
    152 187
    153 
    -    def update_execute_stage(self, stage):
    154 
    -        self._execute_stage = stage
    155 
    -        for queue in self._operation_update_queues:
    156 
    -            queue.put(self.get_operation())
    188 
    +            self._lease.status.Clear()
    189 
    +            self._lease.result.Clear()
    157 190
    158 
    -    def _pack_any(self, pack):
    159 
    -        some_any = any_pb2.Any()
    160 
    -        some_any.Pack(pack)
    161 
    -        return some_any
    191 
    +        elif self._lease.state == LeaseState.ACTIVE.value:
    192 
    +            self.__worker_start_timestamp.GetCurrentTime()
    193 
    +
    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_result.execution_metadata.queued_timestamp.CopyFrom(self.__worker_start_timestamp)
    208 
    +            action_result.execution_metadata.worker_start_timestamp.CopyFrom(self.__worker_start_timestamp)
    209 
    +            action_result.execution_metadata.worker_completed_timestamp.CopyFrom(self.__worker_completed_timestamp)
    210 
    +
    211 
    +            self.__execute_response = remote_execution_pb2.ExecuteResponse()
    212 
    +            self.__execute_response.result.CopyFrom(action_result)
    213 
    +            self.__execute_response.cached_result = False
    214 
    +            self.__execute_response.status.CopyFrom(status)
    215 
    +
    216 
    +    def update_operation_stage(self, stage):
    217 
    +        """Operates a stage transition for the job's :class:Operation.
    218 
    +
    219 
    +        Args:
    220 
    +            stage (OperationStage): the operation stage to transition to.
    221 
    +        """
    222 
    +        if stage.value == self.__operation_metadata.stage:
    223 
    +            return
    224 
    +
    225 
    +        self.__operation_metadata.stage = stage.value
    226 
    +
    227 
    +        if self.__operation_metadata.stage == OperationStage.QUEUED.value:
    228 
    +            if not self.__queued_timestamp:
    229 
    +                self.__queued_timestamp.GetCurrentTime()
    230 
    +            self._n_tries += 1
    231 
    +
    232 
    +        elif self.__operation_metadata.stage == OperationStage.COMPLETED.value:
    233 
    +            if self.__execute_response is not None:
    234 
    +                self._operation.response.Pack(self.__execute_response)
    235 
    +            self._operation.done = True
    236 
    +
    237 
    +        self._operation.metadata.Pack(self.__operation_metadata)
    238 
    +
    239 
    +        for queue in self._operation_update_queues:
    240 
    +            queue.put(self._operation)

  • buildgrid/server/operations/instance.py
    ... ... @@ -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:

  • buildgrid/server/scheduler.py
    ... ... @@ -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
    ... ... @@ -24,10 +22,8 @@ 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 
    -from .job import ExecuteStage, LeaseState
    26 
    +from .job import OperationStage, LeaseState
    31 27
    32 28
    33 29 
     class Scheduler:
    ... ... @@ -39,80 +35,96 @@ 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)
    40 
    +
    41 
    +    def unregister_client(self, job_name, queue):
    42 
    +        self.jobs[job_name].unregister_client(queue)
    44 43
    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]
    44 
    +        if not self.jobs[job_name].n_clients and self.jobs[job_name].operation.done:
    45 
    +            del self.jobs[job_name]
    50 46
    51 
    -    def append_job(self, job, skip_cache_lookup=False):
    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_execute_stage(ExecuteStage.QUEUED)
    59 57
    60 58 
                 else:
    61 
    -                job.result = cached_result
    62 
    -                job.result_cached = True
    63 
    -                job.update_execute_stage(ExecuteStage.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_execute_stage(ExecuteStage.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 
    -                job.update_execute_stage(ExecuteStage.COMPLETED)
    73 
    +                job.update_operation_stage(OperationStage.COMPLETED)
    75 74 
                     # TODO: Mark these jobs as done
    76 75 
                 else:
    77 
    -                job.update_execute_stage(ExecuteStage.QUEUED)
    78 
    -                job.n_tries += 1
    76 
    +                job.update_operation_stage(OperationStage.QUEUED)
    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_execute_stage(ExecuteStage.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()
    114 
    -            job.update_execute_stage(ExecuteStage.EXECUTING)
    115 
    -            job.create_lease()
    116 
    -            job.lease.state = LeaseState.PENDING.value
    117 
    -            return job.lease
    118 
    -        return None
    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.COMPLETED:
    113 
    +            job.update_lease_state(lease_state)
    114 
    +
    115 
    +        else:
    116 
    +            job.update_lease_state(lease_state,
    117 
    +                                   status=lease_status, result=lease_result)
    118 
    +
    119 
    +            if self._action_cache is not None and not job.do_not_cache:
    120 
    +                self._action_cache.update_action_result(job.action_digest, job.action_result)
    121 
    +
    122 
    +            job.update_operation_stage(OperationStage.COMPLETED)
    123 
    +
    124 
    +    def get_job_lease(self, job_name):
    125 
    +        """Returns the lease associated to job, if any have been emitted yet."""
    126 
    +        return self.jobs[job_name].lease
    127 
    +
    128 
    +    def get_job_operation(self, job_name):
    129 
    +        """Returns the operation associated to job."""
    130 
    +        return self.jobs[job_name].operation

  • docs/source/index.rst
    ... ... @@ -19,6 +19,7 @@ Remote execution service implementing Google's REAPI and RWAPI.
    19 19 
        using.rst
    20 20 
        reference.rst
    21 21 
        contributing.rst
    22 
    +   resources.rst
    22 23
    23 24
    24 25 
     Resources
    ... ... @@ -28,11 +29,11 @@ Resources
    28 29 
     - `GitLab repository`_
    29 30 
     - `Bug tracking`_
    30 31 
     - `Mailing list`_
    31 
    -- `Slack channel`_  [`invite link`_]
    32 
    +- `Slack channel`_ [`invite link`_]
    32 33
    33 34 
     .. _Homepage: https://buildgrid.build
    34 35 
     .. _GitLab repository: https://gitlab.com/BuildGrid/buildgrid
    35 
    -.. _Bug tracking: https://gitlab.com/BuildGrid/buildgrid/issues
    36 
    +.. _Bug tracking: https://gitlab.com/BuildGrid/buildgrid/boards
    36 37 
     .. _Mailing list: https://lists.buildgrid.build/cgi-bin/mailman/listinfo/buildgrid
    37 38 
     .. _Slack channel: https://buildteamworld.slack.com/messages/CC9MKC203
    38 
    -.. _invite link: https://join.slack.com/t/buildteamworld/shared_invite/enQtMzkxNzE0MDMyMDY1LTRmZmM1OWE0OTFkMGE1YjU5Njc4ODEzYjc0MGMyOTM5ZTQ5MmE2YTQ1MzQwZDc5MWNhODY1ZmRkZTE4YjFhNjU
    39 
    +.. _invite link: https://join.slack.com/t/buildteamworld/shared_invite/enQtMzkxNzE0MDMyMDY1LTRmZmM1OWE0OTFkMGE1YjU5Njc4ODEzYjc0MGMyOTM5ZTQ5MmE2YTQ1MzQwZDc5MWNhODY1ZmRkZTE4YjFhNjU
    \ No newline at end of file

  • docs/source/installation.rst
    1 
    -
    2 1 
     .. _installation:
    3 2
    4 3 
     Installation
    5 4 
     ============
    6 5
    7 
    -How to install BuildGrid onto your machine.
    6 
    +.. _install-on-host:
    7 
    +
    8 
    +Installation onto host machine
    9 
    +------------------------------
    10 
    +
    11 
    +How to install BuildGrid directly onto your machine.
    8 12
    9 13 
     .. note::
    10 14
    11 
    -   BuildGrid server currently only support *Linux*, *macOS* and *Windows*
    15 
    +   BuildGrid server currently only support *Linux*. *macOS* and *Windows*
    12 16 
        platforms are **not** supported.
    13 17
    14 18
    15 
    -.. _install-prerequisites:
    19 
    +.. _install-host-prerequisites:
    16 20
    17 21 
     Prerequisites
    18 
    --------------
    22 
    +~~~~~~~~~~~~~
    19 23
    20 24 
     BuildGrid only supports ``python3 >= 3.5`` but has no system requirements. Main
    21 
    -Python dependencies, automatically handle during installation, includes:
    25 
    +Python dependencies, automatically handled during installation, include:
    22 26
    23 27 
     - `boto3`_: the Amazon Web Services (AWS) SDK for Python.
    24 28 
     - `click`_: a Python composable command line library.
    ... ... @@ -33,10 +37,10 @@ Python dependencies, automatically handle during installation, includes:
    33 37 
     .. _protocol-buffers: https://developers.google.com/protocol-buffers
    34 38
    35 39
    36 
    -.. _source-install:
    40 
    +.. _install-host-source-install:
    37 41
    38 42 
     Install from sources
    39 
    ---------------------
    43 
    +~~~~~~~~~~~~~~~~~~~~
    40 44
    41 45 
     BuildGrid has ``setuptools`` support. In order to install it to your home
    42 46 
     directory, typically under ``~/.local``, simply run:
    ... ... @@ -46,7 +50,7 @@ directory, typically under ``~/.local``, simply run:
    46 50 
        git clone https://gitlab.com/BuildGrid/buildgrid.git && cd buildgrid
    47 51 
        pip3 install --user --editable .
    48 52
    49 
    -Additionally, and if your distribution does not already includes it, you may
    53 
    +Additionally, and if your distribution does not already include it, you may
    50 54 
     have to adjust your ``PATH``, in ``~/.bashrc``, with:
    51 55
    52 56 
     .. code-block:: sh
    ... ... @@ -63,3 +67,62 @@ have to adjust your ``PATH``, in ``~/.bashrc``, with:
    63 67 
        .. code-block:: sh
    64 68
    65 69 
           pip3 install --user --editable ".[docs,tests]"
    70 
    +
    71 
    +
    72 
    +
    73 
    +.. install-docker:
    74 
    +
    75 
    +Installation through docker
    76 
    +---------------------------
    77 
    +
    78 
    +How to build a Docker image that runs BuildGrid.
    79 
    +
    80 
    +.. _install-docker-prerequisites:
    81 
    +
    82 
    +Prerequisites
    83 
    +~~~~~~~~~~~~~
    84 
    +
    85 
    +A working Docker installation. Please consult `Docker's Getting Started Guide`_ if you don't already have it installed.
    86 
    +
    87 
    +.. _`Docker's Getting Started Guide`: https://www.docker.com/get-started
    88 
    +
    89 
    +
    90 
    +.. _install-docker-build:
    91 
    +
    92 
    +Docker Container from Sources
    93 
    +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    94 
    +
    95 
    +To clone the source code and build a Docker image, simply run:
    96 
    +
    97 
    +.. code-block:: sh
    98 
    +
    99 
    +   git clone https://gitlab.com/BuildGrid/buildgrid.git && cd buildgrid
    100 
    +   docker build -t buildgrid_server .
    101 
    +
    102 
    +.. note::
    103 
    +
    104 
    +   The image built will contain the contents of the source code directory, including
    105 
    +   configuration files.
    106 
    +
    107 
    +.. hint::
    108 
    +
    109 
    +    Whenever the source code is updated or new configuration files are made, you need to re-build
    110 
    +    the image.
    111 
    +
    112 
    +After building the Docker image, to run BuildGrid using the default configuration file
    113 
    +(found in `buildgrid/_app/settings/default.yml`), simply run:
    114 
    +
    115 
    +.. code-block:: sh
    116 
    +
    117 
    +   docker run -i -p 50051:50051 buildgrid_server
    118 
    +
    119 
    +.. note::
    120 
    +
    121 
    +    To run BuildGrid using a different configuration file, include the relative path to the
    122 
    +    configuration file at the end of the command above. For example, to run the default
    123 
    +    standalone CAS server (without an execution service), simply run:
    124 
    +
    125 
    +       .. code-block:: sh
    126 
    +
    127 
    +            docker run -i -p 50052:50052 buildgrid_server buildgrid/_app/settings/cas.yml
    128 
    +

  • docs/source/resources.rst
    1 
    +.. _external-resources:
    2 
    +
    3 
    +Resources
    4 
    +=========
    5 
    +
    6 
    +Remote execution and worker API useful links:
    7 
    +
    8 
    +- `REAPI design document`_
    9 
    +- `REAPI protobuf specification`_
    10 
    +- `RWAPI design document`_
    11 
    +- `RWAPI protobuf specification`_
    12 
    +- `Bazel`_ `remote caching and execution documentation`_
    13 
    +- `RECC usage instructions`_
    14 
    +- `BuildStream webside`_ and `documentation`_
    15 
    +- `BuildStream-externals repository`_
    16 
    +- `BuildBox repository`_
    17 
    +- `FUSE on wikipedia`_ and `kernel documentation`_
    18 
    +- `bubblewrap repository`_
    19 
    +- `Buildfarm reference REAPI implementation`_
    20 
    +- `Buildbarn Golang REAPI implementation`_
    21 
    +- `Demonstration of RECC with BuildGrid`_
    22 
    +- `Demonstration of Bazel with BuildGrid`_
    23 
    +- `Demonstration of BuildStream with BuildGrid`_
    24 
    +
    25 
    +.. _REAPI design document: https://docs.google.com/document/d/1AaGk7fOPByEvpAbqeXIyE8HX_A3_axxNnvroblTZ_6s
    26 
    +.. _REAPI protobuf specification: https://github.com/bazelbuild/remote-apis/blob/master/build/bazel/remote/execution/v2/remote_execution.proto
    27 
    +.. _RWAPI design document: https://docs.google.com/document/d/1s_AzRRD2mdyktKUj2HWBn99rMg_3tcPvdjx3MPbFidU
    28 
    +.. _RWAPI protobuf specification: https://github.com/googleapis/googleapis/blob/master/google/devtools/remoteworkers/v1test2/bots.proto
    29 
    +.. _Bazel: https://www.bazel.build
    30 
    +.. _remote caching and execution documentation: https://docs.bazel.build/versions/master/remote-caching.html
    31 
    +.. _RECC usage instructions: https://gitlab.com/bloomberg/recc#running-recc
    32 
    +.. _BuildStream webside: https://buildstream.build
    33 
    +.. _documentation: https://docs.buildstream.build
    34 
    +.. _BuildStream-externals repository: https://gitlab.com/BuildStream/bst-external
    35 
    +.. _FUSE on wikipedia: https://en.wikipedia.org/wiki/Filesystem_in_Userspace
    36 
    +.. _kernel documentation: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/filesystems/fuse.txt
    37 
    +.. _BuildBox repository: https://gitlab.com/BuildStream/buildbox
    38 
    +.. _bubblewrap repository: https://github.com/projectatomic/bubblewrap
    39 
    +.. _Buildfarm reference REAPI implementation: https://github.com/bazelbuild/bazel-buildfarm
    40 
    +.. _Buildbarn Golang REAPI implementation: https://github.com/EdSchouten/bazel-buildbarn
    41 
    +.. _Demonstration of RECC with BuildGrid: https://asciinema.org/a/0FjExIqrTGSlpSUIS8Ehf5gUg
    42 
    +.. _Demonstration of Bazel with BuildGrid: https://asciinema.org/a/uVHFWOxpivwJ4ari23CEerR8N
    43 
    +.. _Demonstration of BuildStream with BuildGrid: https://asciinema.org/a/QfkYGqhfhEQz4o8prlBdEBFP7

  • setup.cfg
    ... ... @@ -14,3 +14,6 @@ pep8ignore =
    14 14 
         docs/source/conf.py ALL
    15 15 
         *_pb2.py ALL
    16 16 
         *_pb2_grpc.py ALL
    17 
    +filterwarnings =
    18 
    +    ignore::DeprecationWarning
    19 
    +    ignore::PendingDeprecationWarning
    \ No newline at end of file

  • setup.py
    ... ... @@ -90,7 +90,7 @@ tests_require = [
    90 90 
         'moto',
    91 91 
         'pep8',
    92 92 
         'psutil',
    93 
    -    'pytest == 3.6.4',
    93 
    +    'pytest >= 3.8.0',
    94 94 
         'pytest-cov >= 2.6.0',
    95 95 
         'pytest-pep8',
    96 96 
         'pytest-pylint',

  • tests/integration/bots_service.py
    ... ... @@ -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)

  • tests/integration/execution_service.py
    ... ... @@ -82,7 +82,7 @@ def test_execute(skip_cache_lookup, instance, context):
    82 82 
         assert isinstance(result, operations_pb2.Operation)
    83 83 
         metadata = remote_execution_pb2.ExecuteOperationMetadata()
    84 84 
         result.metadata.Unpack(metadata)
    85 
    -    assert metadata.stage == job.ExecuteStage.QUEUED.value
    85 
    +    assert metadata.stage == job.OperationStage.QUEUED.value
    86 86 
         assert uuid.UUID(result.name, version=4)
    87 87 
         assert result.done is False
    88 88
    ... ... @@ -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))
    ... ... @@ -116,7 +116,7 @@ def test_wait_execution(instance, controller, context):
    116 116 
         action_result = remote_execution_pb2.ActionResult()
    117 117 
         action_result_any.Pack(action_result)
    118 118
    119 
    -    j.update_execute_stage(job.ExecuteStage.COMPLETED)
    119 
    +    j.update_operation_stage(job.OperationStage.COMPLETED)
    120 120
    121 121 
         response = instance.WaitExecution(request, context)
    122 122
    ... ... @@ -125,7 +125,7 @@ def test_wait_execution(instance, controller, context):
    125 125 
         assert isinstance(result, operations_pb2.Operation)
    126 126 
         metadata = remote_execution_pb2.ExecuteOperationMetadata()
    127 127 
         result.metadata.Unpack(metadata)
    128 
    -    assert metadata.stage == job.ExecuteStage.COMPLETED.value
    128 
    +    assert metadata.stage == job.OperationStage.COMPLETED.value
    129 129 
         assert uuid.UUID(result.name, version=4)
    130 130 
         assert result.done is True
    131 131

  • tests/integration/operations_service.py
    ... ... @@ -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)

    • —
    View it on GitLab.
    You're receiving this email because of your account on gitlab.com. If you'd like to receive fewer emails, you can adjust your notification settings.

    [Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]