Skip to content

Ingester

Sequence Diagram

sequenceDiagram
    Ingester ->>+ accept_workers: run
    Ingester ->>+ work: run
    Ingester ->>+ manage_assignments: run
    Ingester ->>+ update_metrics: run
    Ingester ->>+ register: run


    work ->>- Ingester: restart work
    manage_assignments ->>- Ingester: restart work
    Ingester ->>+ work: restart work
    Ingester ->>+ manage_assignments: restart work

    accept_workers ->>- Ingester: close
    work ->>- Ingester: close
    update_metrics ->>- Ingester: close
    manage_assignments ->>- Ingester: close

Ingester

Bases: DistributedService

The Ingester class provides a basis to write custom ingesters for any protocol. It handles all the forwarding to workers and managing assignments. For the instance to do anything, await run()

Source code in dranspose/ingester.py 
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
class Ingester(DistributedService):
    """
    The Ingester class provides a basis to write custom ingesters for any protocol.
    It handles all the forwarding to workers and managing assignments.
    For the instance to do anything, await `run()`
    """

    def __init__(self, settings: Optional[IngesterSettings] = None):
        if settings is None:
            settings = IngesterSettings()
        self._ingester_settings = settings
        state = IngesterState(
            name=self._ingester_settings.ingester_name,
            url=self._ingester_settings.ingester_url,
            streams=self._ingester_settings.ingester_streams,
        )

        super().__init__(state=state, settings=self._ingester_settings)
        self._logger.info(
            "created ingester with state %s and settings %s",
            state,
            self._ingester_settings,
        )
        self.state: IngesterState

        self.dump_file: Optional[BufferedWriter] = None

    def open_socket(self) -> None:
        self.ctx = zmq.asyncio.Context()
        self.out_socket = self.ctx.socket(zmq.ROUTER)
        self.out_socket.setsockopt(zmq.ROUTER_MANDATORY, 1)
        self.out_socket.setsockopt(zmq.TCP_KEEPALIVE, 1)
        self.out_socket.setsockopt(zmq.TCP_KEEPALIVE_IDLE, 300)
        self.out_socket.setsockopt(zmq.TCP_KEEPALIVE_INTVL, 300)
        self.out_socket.bind(f"tcp://*:{self._ingester_settings.ingester_url.port}")

    async def run(self) -> None:
        """
        Main function orchestrating the dependent tasks. This needs to be called from an async context once an instance is created.
        """
        self.open_socket()

        self.accept_task = asyncio.create_task(self.accept_workers())
        self.accept_task.add_done_callback(done_callback)
        self.work_task = asyncio.create_task(self.work())
        self.work_task.add_done_callback(done_callback)
        self.assign_task = asyncio.create_task(self.manage_assignments())
        self.assign_task.add_done_callback(done_callback)
        self.assignment_queue: asyncio.Queue[WorkAssignment] = asyncio.Queue()
        self.metrics_task = asyncio.create_task(self.update_metrics())
        self.metrics_task.add_done_callback(done_callback)
        self._logger.info("all subtasks running")
        await self.register()

    async def restart_work(self, new_uuid: UUID4) -> None:
        """
        Restarts all work related tasks to make sure no old state is present in a new scan.

        Arguments:
            new_uuid: The uuid of the new mapping
        """
        await cancel_and_wait(self.work_task)
        await cancel_and_wait(self.assign_task)
        self.state.mapping_uuid = new_uuid
        self.assignment_queue = asyncio.Queue()
        self.work_task = asyncio.create_task(self.work())
        self.work_task.add_done_callback(done_callback)
        self.assign_task = asyncio.create_task(self.manage_assignments())
        self.assign_task.add_done_callback(done_callback)

    async def finish_work(self) -> None:
        """
        This hook is called when all events of a trigger map were ingested. It is useful for e.g. closing open files.
        """
        self._logger.info("finishing work")
        if self.dump_file:
            self.dump_file.close()
            self._logger.info("closed dump file at finish %s", self.dump_file)
            self.dump_file = None

        await self.redis.xadd(
            RedisKeys.ready(self.state.mapping_uuid),
            {
                "data": IngesterUpdate(
                    state=DistributedStateEnum.FINISHED,
                    ingester=self.state.name,
                ).model_dump_json()
            },
        )

    async def manage_assignments(self) -> None:
        """
        A background task reading assignments from the controller, filering them for the relevant ones and enqueueing them for when the frame arrives.
        """
        self._logger.info("started ingester manage assign task")
        lastev = 0
        while True:
            sub = RedisKeys.assigned(self.state.mapping_uuid)
            try:
                assignments = await self.redis.xread({sub: lastev}, block=1000)
            except rexceptions.ConnectionError:
                break
            if sub not in assignments:
                continue
            assignment_evs = assignments[sub][0]
            self._logger.debug("got assignments %s", assignment_evs)
            for assignment in assignment_evs:
                was = WorkAssignmentList.validate_json(assignment[1]["data"])
                for wa in was:
                    mywa = wa.get_workers_for_streams(self.state.streams)
                    if len(mywa.assignments) > 0:
                        await self.assignment_queue.put(mywa)
                lastev = assignment[0]

    async def _send_workermessages(
        self, workermessages: dict[WorkerName, InternalWorkerMessage]
    ) -> None:
        for worker, message in workermessages.items():
            self._logger.debug(
                "header is %s",
                message.model_dump_json(exclude={"streams": {"__all__": "frames"}}),
            )
            await self.out_socket.send_multipart(
                [worker.encode("ascii")]
                + [
                    message.model_dump_json(
                        exclude={"streams": {"__all__": "frames"}}
                    ).encode("utf8")
                ]
                + message.get_all_frames()
            )
            self._logger.debug("sent message to worker %s", worker)

    async def work(self) -> None:
        """
        The heavy liftig function of an ingester. It consumes a generator `run_source()` which
        should be implemented for a specific protocol.
        It then assembles all streams for this ingester and forwards them to the assigned workers.

        Optionally the worker dumps the internal messages to disk. This is useful to develop workers with actual data captured.
        """
        self._logger.info("started ingester work task")
        sourcegens = {stream: self.run_source(stream) for stream in self.state.streams}
        self.dump_file = None
        self.dump_filename = self._ingester_settings.dump_path
        if self.dump_filename is None and "dump_prefix" in self.parameters:
            val = self.parameters[ParameterName("dump_prefix")].data.decode("utf8")
            if len(val) > 0:
                self.dump_filename = f"{val}{self._ingester_settings.ingester_name}-{self.state.mapping_uuid}.cbors"
        if self.dump_filename:
            self.dump_file = open(self.dump_filename, "ab")
            self._logger.info(
                "dump file %s opened at %s",
                self.dump_filename,
                self.dump_file,
            )
        try:
            took = []
            empties = []
            while True:
                empty = False
                if self.assignment_queue.empty():
                    empty = True
                start = time.perf_counter()
                work_assignment: WorkAssignment = await self.assignment_queue.get()
                if empty:
                    empties.append(work_assignment.event_number)

                workermessages: dict[WorkerName, InternalWorkerMessage] = {}
                zmqyields: list[Awaitable[StreamData]] = []
                streams: list[StreamName] = []
                for stream in work_assignment.assignments:
                    zmqyields.append(anext(sourcegens[stream]))
                    streams.append(stream)
                try:
                    zmqstreams: list[StreamData] = await asyncio.gather(*zmqyields)
                except StopAsyncIteration:
                    self._logger.warning("stream source stopped before end")
                    raise asyncio.exceptions.CancelledError()
                zmqparts: dict[StreamName, StreamData] = {
                    stream: zmqpart for stream, zmqpart in zip(streams, zmqstreams)
                }
                if self.dump_file:
                    self._logger.debug("writing dump to path %s", self.dump_filename)
                    allstr = InternalWorkerMessage(
                        event_number=work_assignment.event_number,
                        streams={k: v.get_bytes() for k, v in zmqparts.items()},
                    )
                    try:
                        cbor2.dump(
                            CBORTag(55799, allstr),
                            self.dump_file,
                            default=message_encoder,
                        )
                    except Exception as e:
                        self._logger.error("cound not dump %s", e.__repr__())
                    self._logger.debug("written dump")
                for stream, workers in work_assignment.assignments.items():
                    for worker in workers:
                        if worker not in workermessages:
                            workermessages[worker] = InternalWorkerMessage(
                                event_number=work_assignment.event_number
                            )
                        workermessages[worker].streams[stream] = zmqparts[stream]
                self._logger.debug("workermessages %s", workermessages)
                await self._send_workermessages(workermessages)
                end = time.perf_counter()
                took.append(end - start)
                if len(took) > 1000:
                    self._logger.info(
                        "forwarding took avg %lf, min %f max %f",
                        sum(took) / len(took),
                        min(took),
                        max(took),
                    )
                    self._logger.info("waiting for queue %d of 1000", len(empties))
                    took = []
                    empties = []
                self.state.processed_events += 1
        except asyncio.exceptions.CancelledError:
            self._logger.info("stopping worker")
            for stream in self.state.streams:
                await self.stop_source(stream)
            if self.dump_file:
                self._logger.info(
                    "closing dump file %s at cancelled work", self.dump_file
                )
                self.dump_file.close()
                self.dump_file = None

    async def run_source(self, stream: StreamName) -> AsyncGenerator[StreamData, None]:
        """
        This generator must be implemented by the customised subclass. It should return exactly one `StreamData` object
        for every frame arriving from upstream.

        Arguments:
            stream: optionally it received a stream name for which is should yield frames.

        Returns:
            Yield a StreamData object for every received frame.
        """
        yield StreamData(typ="", frames=[])
        return

    async def stop_source(self, stream: StreamName) -> None:
        pass

    async def accept_workers(self) -> None:
        """
        To allow zmq to learn the names of attached workers, they periodically send empty packets.
        There is no information flow directly from workers to ingesters, so we discard the data.
        """
        poller = zmq.asyncio.Poller()
        poller.register(self.out_socket, zmq.POLLIN)
        while True:
            socks = dict(await poller.poll())
            for sock in socks:
                data = await sock.recv_multipart()
                connected_worker = ConnectedWorker(
                    name=data[0], service_uuid=UUID(bytes=data[1])
                )
                fast_publish = False
                if connected_worker.service_uuid not in self.state.connected_workers:
                    fast_publish = True
                self.state.connected_workers[
                    connected_worker.service_uuid
                ] = connected_worker
                now = time.time()
                self.state.connected_workers = {
                    uuid: cw
                    for uuid, cw in self.state.connected_workers.items()
                    if now - cw.last_seen < 3
                }
                self._logger.debug("worker pinnged %s", connected_worker)
                if fast_publish:
                    self._logger.debug("fast publish")
                    await self.publish_config()

    async def close(self) -> None:
        """
        Clean up any open connections
        """
        await cancel_and_wait(self.accept_task)
        await cancel_and_wait(self.work_task)
        await cancel_and_wait(self.metrics_task)
        await cancel_and_wait(self.assign_task)
        await self.redis.delete(RedisKeys.config("ingester", self.state.name))
        await super().close()
        self.ctx.destroy(linger=0)
        self._logger.info("closed ingester")

accept_workers() async

To allow zmq to learn the names of attached workers, they periodically send empty packets. There is no information flow directly from workers to ingesters, so we discard the data.

Source code in dranspose/ingester.py 
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
async def accept_workers(self) -> None:
    """
    To allow zmq to learn the names of attached workers, they periodically send empty packets.
    There is no information flow directly from workers to ingesters, so we discard the data.
    """
    poller = zmq.asyncio.Poller()
    poller.register(self.out_socket, zmq.POLLIN)
    while True:
        socks = dict(await poller.poll())
        for sock in socks:
            data = await sock.recv_multipart()
            connected_worker = ConnectedWorker(
                name=data[0], service_uuid=UUID(bytes=data[1])
            )
            fast_publish = False
            if connected_worker.service_uuid not in self.state.connected_workers:
                fast_publish = True
            self.state.connected_workers[
                connected_worker.service_uuid
            ] = connected_worker
            now = time.time()
            self.state.connected_workers = {
                uuid: cw
                for uuid, cw in self.state.connected_workers.items()
                if now - cw.last_seen < 3
            }
            self._logger.debug("worker pinnged %s", connected_worker)
            if fast_publish:
                self._logger.debug("fast publish")
                await self.publish_config()

close() async

Clean up any open connections

Source code in dranspose/ingester.py 
330
331
332
333
334
335
336
337
338
339
340
341
async def close(self) -> None:
    """
    Clean up any open connections
    """
    await cancel_and_wait(self.accept_task)
    await cancel_and_wait(self.work_task)
    await cancel_and_wait(self.metrics_task)
    await cancel_and_wait(self.assign_task)
    await self.redis.delete(RedisKeys.config("ingester", self.state.name))
    await super().close()
    self.ctx.destroy(linger=0)
    self._logger.info("closed ingester")

finish_work() async

This hook is called when all events of a trigger map were ingested. It is useful for e.g. closing open files.

Source code in dranspose/ingester.py 
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
async def finish_work(self) -> None:
    """
    This hook is called when all events of a trigger map were ingested. It is useful for e.g. closing open files.
    """
    self._logger.info("finishing work")
    if self.dump_file:
        self.dump_file.close()
        self._logger.info("closed dump file at finish %s", self.dump_file)
        self.dump_file = None

    await self.redis.xadd(
        RedisKeys.ready(self.state.mapping_uuid),
        {
            "data": IngesterUpdate(
                state=DistributedStateEnum.FINISHED,
                ingester=self.state.name,
            ).model_dump_json()
        },
    )

manage_assignments() async

A background task reading assignments from the controller, filering them for the relevant ones and enqueueing them for when the frame arrives.

Source code in dranspose/ingester.py 
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
async def manage_assignments(self) -> None:
    """
    A background task reading assignments from the controller, filering them for the relevant ones and enqueueing them for when the frame arrives.
    """
    self._logger.info("started ingester manage assign task")
    lastev = 0
    while True:
        sub = RedisKeys.assigned(self.state.mapping_uuid)
        try:
            assignments = await self.redis.xread({sub: lastev}, block=1000)
        except rexceptions.ConnectionError:
            break
        if sub not in assignments:
            continue
        assignment_evs = assignments[sub][0]
        self._logger.debug("got assignments %s", assignment_evs)
        for assignment in assignment_evs:
            was = WorkAssignmentList.validate_json(assignment[1]["data"])
            for wa in was:
                mywa = wa.get_workers_for_streams(self.state.streams)
                if len(mywa.assignments) > 0:
                    await self.assignment_queue.put(mywa)
            lastev = assignment[0]

restart_work(new_uuid) async

Restarts all work related tasks to make sure no old state is present in a new scan.

Parameters:

Name Type Description Default
new_uuid UUID4

The uuid of the new mapping

 required
Source code in dranspose/ingester.py 
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
async def restart_work(self, new_uuid: UUID4) -> None:
    """
    Restarts all work related tasks to make sure no old state is present in a new scan.

    Arguments:
        new_uuid: The uuid of the new mapping
    """
    await cancel_and_wait(self.work_task)
    await cancel_and_wait(self.assign_task)
    self.state.mapping_uuid = new_uuid
    self.assignment_queue = asyncio.Queue()
    self.work_task = asyncio.create_task(self.work())
    self.work_task.add_done_callback(done_callback)
    self.assign_task = asyncio.create_task(self.manage_assignments())
    self.assign_task.add_done_callback(done_callback)

run() async

Main function orchestrating the dependent tasks. This needs to be called from an async context once an instance is created.

Source code in dranspose/ingester.py 
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
async def run(self) -> None:
    """
    Main function orchestrating the dependent tasks. This needs to be called from an async context once an instance is created.
    """
    self.open_socket()

    self.accept_task = asyncio.create_task(self.accept_workers())
    self.accept_task.add_done_callback(done_callback)
    self.work_task = asyncio.create_task(self.work())
    self.work_task.add_done_callback(done_callback)
    self.assign_task = asyncio.create_task(self.manage_assignments())
    self.assign_task.add_done_callback(done_callback)
    self.assignment_queue: asyncio.Queue[WorkAssignment] = asyncio.Queue()
    self.metrics_task = asyncio.create_task(self.update_metrics())
    self.metrics_task.add_done_callback(done_callback)
    self._logger.info("all subtasks running")
    await self.register()

run_source(stream) async

This generator must be implemented by the customised subclass. It should return exactly one StreamData object for every frame arriving from upstream.

Parameters:

Name Type Description Default
stream StreamName

optionally it received a stream name for which is should yield frames.

 required

Returns:

Type Description
AsyncGenerator[StreamData, None]

Yield a StreamData object for every received frame.
Source code in dranspose/ingester.py 
282
283
284
285
286
287
288
289
290
291
292
293
294
async def run_source(self, stream: StreamName) -> AsyncGenerator[StreamData, None]:
    """
    This generator must be implemented by the customised subclass. It should return exactly one `StreamData` object
    for every frame arriving from upstream.

    Arguments:
        stream: optionally it received a stream name for which is should yield frames.

    Returns:
        Yield a StreamData object for every received frame.
    """
    yield StreamData(typ="", frames=[])
    return

work() async

The heavy liftig function of an ingester. It consumes a generator run_source() which should be implemented for a specific protocol. It then assembles all streams for this ingester and forwards them to the assigned workers.

Optionally the worker dumps the internal messages to disk. This is useful to develop workers with actual data captured.

Source code in dranspose/ingester.py 
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
async def work(self) -> None:
    """
    The heavy liftig function of an ingester. It consumes a generator `run_source()` which
    should be implemented for a specific protocol.
    It then assembles all streams for this ingester and forwards them to the assigned workers.

    Optionally the worker dumps the internal messages to disk. This is useful to develop workers with actual data captured.
    """
    self._logger.info("started ingester work task")
    sourcegens = {stream: self.run_source(stream) for stream in self.state.streams}
    self.dump_file = None
    self.dump_filename = self._ingester_settings.dump_path
    if self.dump_filename is None and "dump_prefix" in self.parameters:
        val = self.parameters[ParameterName("dump_prefix")].data.decode("utf8")
        if len(val) > 0:
            self.dump_filename = f"{val}{self._ingester_settings.ingester_name}-{self.state.mapping_uuid}.cbors"
    if self.dump_filename:
        self.dump_file = open(self.dump_filename, "ab")
        self._logger.info(
            "dump file %s opened at %s",
            self.dump_filename,
            self.dump_file,
        )
    try:
        took = []
        empties = []
        while True:
            empty = False
            if self.assignment_queue.empty():
                empty = True
            start = time.perf_counter()
            work_assignment: WorkAssignment = await self.assignment_queue.get()
            if empty:
                empties.append(work_assignment.event_number)

            workermessages: dict[WorkerName, InternalWorkerMessage] = {}
            zmqyields: list[Awaitable[StreamData]] = []
            streams: list[StreamName] = []
            for stream in work_assignment.assignments:
                zmqyields.append(anext(sourcegens[stream]))
                streams.append(stream)
            try:
                zmqstreams: list[StreamData] = await asyncio.gather(*zmqyields)
            except StopAsyncIteration:
                self._logger.warning("stream source stopped before end")
                raise asyncio.exceptions.CancelledError()
            zmqparts: dict[StreamName, StreamData] = {
                stream: zmqpart for stream, zmqpart in zip(streams, zmqstreams)
            }
            if self.dump_file:
                self._logger.debug("writing dump to path %s", self.dump_filename)
                allstr = InternalWorkerMessage(
                    event_number=work_assignment.event_number,
                    streams={k: v.get_bytes() for k, v in zmqparts.items()},
                )
                try:
                    cbor2.dump(
                        CBORTag(55799, allstr),
                        self.dump_file,
                        default=message_encoder,
                    )
                except Exception as e:
                    self._logger.error("cound not dump %s", e.__repr__())
                self._logger.debug("written dump")
            for stream, workers in work_assignment.assignments.items():
                for worker in workers:
                    if worker not in workermessages:
                        workermessages[worker] = InternalWorkerMessage(
                            event_number=work_assignment.event_number
                        )
                    workermessages[worker].streams[stream] = zmqparts[stream]
            self._logger.debug("workermessages %s", workermessages)
            await self._send_workermessages(workermessages)
            end = time.perf_counter()
            took.append(end - start)
            if len(took) > 1000:
                self._logger.info(
                    "forwarding took avg %lf, min %f max %f",
                    sum(took) / len(took),
                    min(took),
                    max(took),
                )
                self._logger.info("waiting for queue %d of 1000", len(empties))
                took = []
                empties = []
            self.state.processed_events += 1
    except asyncio.exceptions.CancelledError:
        self._logger.info("stopping worker")
        for stream in self.state.streams:
            await self.stop_source(stream)
        if self.dump_file:
            self._logger.info(
                "closing dump file %s at cancelled work", self.dump_file
            )
            self.dump_file.close()
            self.dump_file = None