GeezerVisonThe World Wide Web, restored to 800 pixels of glory. |
| Home|About|Gaming cabinet|Archives|How it works|Starter sites |
GeezerVison is showing github.com — feldera/python/feldera/pipeline.py at faster-parser · feldera/feldera Expand file tree/ Copy pathpipeline.pyMore file actions 2385 lines (1939 loc) · 92.6 KB / Copy pathpipeline.pyFile metadata and controls2385 lines (1939 loc) · 92.6 KB 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 import logging import pathlib import time import warnings from collections import deque from datetime import datetime from threading import Event from typing import ( Any, Callable, Dict, Generator, List, Mapping, Optional, ) from uuid import UUID import pandas import pyarrow as pa from feldera._callback_runner import CallbackRunner from feldera._helpers import chunk_dataframe, ensure_dataframe_has_columns from feldera._long_operation_warning import LongOperationWarning from feldera.enums import ( BootstrapPolicy, CheckpointStatus, CompletionTokenStatus, DeploymentDesiredStatus, DeploymentResourcesDesiredStatus, DeploymentResourcesStatus, DeploymentRuntimeDesiredStatus, DeploymentRuntimeStatus, PipelineFieldSelector, PipelineStatus, ProgramStatus, StorageStatus, TransactionStatus, ) from feldera.output_handler import OutputHandler from feldera.rest.errors import ( FelderaAPIError, FelderaCommunicationError, FelderaTimeoutError, ) from feldera.rest.feldera_client import FelderaClient from feldera.rest.logs import LogStream from feldera.rest.pipeline import Pipeline as InnerPipeline from feldera.rest.sql_table import SQLTable from feldera.rest.sql_view import SQLView from feldera.runtime_config import RuntimeConfig from feldera.stats import InputEndpointStatus, OutputEndpointStatus, PipelineStatistics from feldera.types import CheckpointMetadata logger = logging.getLogger(__name__) class Pipeline: def __init__(self, client: FelderaClient): self.client: FelderaClient = client self._inner: InnerPipeline | None = None @staticmethod def _from_inner(inner: InnerPipeline, client: FelderaClient) -> "Pipeline": pipeline = Pipeline(client) pipeline._inner = inner return pipeline def refresh(self, field_selector: PipelineFieldSelector): """ Calls the backend to get the updated, latest version of the pipeline. :param field_selector: Choose what pipeline information to refresh; see PipelineFieldSelector enum definition. :raises FelderaConnectionError: If there is an issue connecting to the backend. """ self._inner = self.client.get_pipeline(self.name, field_selector) def status(self) -> PipelineStatus: """ Return the current status of the pipeline. """ try: self.refresh(PipelineFieldSelector.STATUS) return PipelineStatus.from_str(self._inner.deployment_status) except FelderaAPIError as err: if err.status_code == 404: return PipelineStatus.NOT_FOUND else: raise err def wait_for_status( self, expected_status: PipelineStatus, timeout: Optional[int] = None ) -> None: """ Wait for the pipeline to reach the specified status. :param expected_status: The status to wait for :param timeout: Maximum time to wait in seconds. If None, waits forever (default: None) :raises TimeoutError: If the expected status is not reached within the timeout """ start_time = time.monotonic() while True: current_status = self.status() if current_status == expected_status: return if timeout is not None and time.monotonic() - start_time >= timeout: raise TimeoutError( f"Pipeline did not reach {expected_status.name} status within {timeout} seconds" ) time.sleep(1) def stats(self) -> PipelineStatistics: """Gets the pipeline metrics and performance counters.""" return PipelineStatistics.from_dict(self.client.get_pipeline_stats(self.name)) def metrics(self, format: str = "prometheus"): """Gets the pipeline circuit metrics. :param format: The metrics format, "prometheus" (default) or "json". :return: Prometheus exposition text as `str`, or a `dict` for "json". """ return self.client.get_pipeline_metrics(self.name, format=format) def logs(self, follow: bool = True) -> Generator[str, None, None]: """Gets the pipeline logs. :param follow: Set False to stop after the retained logs instead of waiting for new lines. """ return self.client.get_pipeline_logs(self.name, follow=follow) def resume_logs(self, cursor: Optional[str] = None) -> LogStream: """ Open the pipeline logs stream, starting from the last read line (cursor). Where :meth:`logs` replays the whole retained buffer on every connection, this delivers only the lines following `cursor`. The logs buffer is discarded when the runner restarts or the pipeline is deleted, and survives a pipeline being stopped and started. A cursor issued during an earlier lifetime of the buffer is not an error: the stream starts at the beginning of the retained buffer, and `LogPosition.gap` names the lines that were lost. For example, to keep reading the logs across dropped connections:: cursor = None while True: with pipeline.resume_logs(cursor) as stream: for line in stream: print(line) cursor = stream.cursor() :param cursor: The position to resume after, as returned by :meth:`feldera.rest.logs.LogStream.cursor`. `None` starts at the beginning of the retained buffer. :return: The open stream. Close it when done reading. """ return self.client.resume_pipeline_logs(self.name, cursor) def input_pandas( self, table_name: str, df: pandas.DataFrame, force: bool = False, connector_metadata: Optional[Mapping[str, Any]] = None, ) -> Optional[str]: """ Push all rows in a pandas DataFrame to the pipeline. The pipeline must either be in RUNNING or PAUSED states to push data. An error will be raised if the pipeline is in any other state. The dataframe must have the same columns as the table in the pipeline. :param table_name: The name of the table to insert data into. :param df: The pandas DataFrame to be pushed to the pipeline. :param force: `True` to push data even if the pipeline is paused. `False` by default. :param connector_metadata: Connector metadata for every row of the DataFrame, as a dict of attribute names to values. ``CONNECTOR_METADATA()`` returns it for the rows. None by default. :returns: The completion token of the last chunk pushed, which also covers the earlier chunks, or None if the DataFrame is empty and nothing was pushed. :raises ValueError: If the table does not exist in the pipeline. :raises RuntimeError: If the pipeline is not in a valid state to push data. :raises RuntimeError: If the pipeline is paused and force is not set to `True`. """ status = self.status() if status not in [ PipelineStatus.RUNNING, PipelineStatus.PAUSED, PipelineStatus.CONCURRENTBOOTSTRAPPING, PipelineStatus.SYNCHRONIZING, ]: raise RuntimeError("Pipeline must be running or paused to push data") if not force and status == PipelineStatus.PAUSED: raise RuntimeError("Pipeline is paused, set force=True to push data") ensure_dataframe_has_columns(df) pipeline = self.client.get_pipeline(self.name, PipelineFieldSelector.ALL) if table_name.lower() != "now" and table_name.lower() not in [ tbl.name.lower() for tbl in pipeline.tables ]: raise ValueError( f"Cannot push to table '{table_name}': table with this name" f" does not exist in the '{self.name}' pipeline" ) else: # consider validating the schema here token = None for datum in chunk_dataframe(df): token = self.client.push_to_pipeline( self.name, table_name, "json", datum.to_json(orient="records", date_format="epoch"), json_flavor="pandas", array=True, serialize=False, force=force, connector_metadata=connector_metadata, ) return token def input_json( self, table_name: str, data: Dict | list, update_format: str = "raw", force: bool = False, wait: bool = True, connector_metadata: Optional[Mapping[str, Any]] = None, **kwargs, ) -> str: """ Push this JSON data to the specified table of the pipeline. The pipeline must either be in RUNNING or PAUSED states to push data. An error will be raised if the pipeline is in any other state. :param table_name: The name of the table to push data into. :param data: The JSON encoded data to be pushed to the pipeline. The data should be in the form: `{'col1': 'val1', 'col2': 'val2'}` or `[{'col1': 'val1', 'col2': 'val2'}, {'col1': 'val1', 'col2': 'val2'}]` :param update_format: The update format of the JSON data to be pushed to the pipeline. Must be one of: "raw", "insert_delete". https://docs.feldera.com/formats/json#the-insertdelete-format :param force: `True` to push data even if the pipeline is paused. `False` by default. :param wait: If True, blocks until this input has been processed by the pipeline :param connector_metadata: Connector metadata for every record of this request, as a dict of attribute names to values. ``CONNECTOR_METADATA()`` returns it for the records, so a column declared with ``DEFAULT CAST(CONNECTOR_METADATA()['name'] AS type)`` takes its value from the ``name`` attribute. None by default. :param kwargs: Additional key word arguments forwarded to the Client push_to_pipeline method :returns: The completion token to this input. :raises ValueError: If the update format is invalid. :raises FelderaAPIError: If the pipeline is not in a valid state to push data. :raises RuntimeError: If the pipeline is paused and `force` is not set to `True`. """ status = self.status() if not force and status == PipelineStatus.PAUSED: raise RuntimeError("Pipeline is paused, set force=True to push data") if update_format not in ["raw", "insert_delete"]: raise ValueError("update_format must be one of raw or insert_delete") array = True if isinstance(data, list) else False return self.client.push_to_pipeline( self.name, table_name, "json", data, update_format=update_format, array=array, force=force, wait=wait, connector_metadata=connector_metadata, **kwargs, ) def pause_input_connector(self, table_name: str, connector_name: str): """ Pause the specified input connector. A paused input connector stops fetching data from its source. All connectors are RUNNING by default. Refer to the connector documentation for more information: https://docs.feldera.com/connectors/orchestration#input-connectors :param table_name: The name of the table that the connector is attached to. :param connector_name: The name of the connector to pause. :raises FelderaAPIError: If the connector is not found, or if the pipeline is not running. """ self.client.pause_input_connector(self.name, table_name, connector_name) def start_input_connector(self, table_name: str, connector_name: str): """ Start (resume) the specified input connector. Refer to the connector documentation for more information: https://docs.feldera.com/connectors/orchestration#input-connectors :param table_name: The name of the table that the connector is attached to. :param connector_name: The name of the connector to start. :raises FelderaAPIError: If the connector is not found, or if the pipeline is not running. """ self.client.start_input_connector(self.name, table_name, connector_name) def pause_connector(self, table_name: str, connector_name: str): """ Pause the specified input connector. .. deprecated:: Use :meth:`.pause_input_connector`, which names the kind of connector it acts on, like :meth:`.pause_output_connector`. """ warnings.warn( "pause_connector is deprecated; use pause_input_connector.", DeprecationWarning, stacklevel=2, ) self.pause_input_connector(table_name, connector_name) def resume_connector(self, table_name: str, connector_name: str): """ Start (resume) the specified input connector. .. deprecated:: Use :meth:`.start_input_connector`, which names the kind of connector it acts on, like :meth:`.start_output_connector`. """ warnings.warn( "resume_connector is deprecated; use start_input_connector.", DeprecationWarning, stacklevel=2, ) self.start_input_connector(table_name, connector_name) def pause_output_connector(self, view_name: str, connector_name: str): """ Pause the specified output connector. A paused output connector discards the output it receives instead of writing it to its sink. Output produced while the connector is paused is gone for good. All connectors are RUNNING by default. Refer to the connector documentation for more information: https://docs.feldera.com/connectors/orchestration#output-connectors :param view_name: The name of the view that the connector is attached to. :param connector_name: The name of the connector to pause. :raises FelderaAPIError: If the connector is not found, or if the pipeline is not running. """ self.client.pause_output_connector(self.name, view_name, connector_name) def start_output_connector(self, view_name: str, connector_name: str): """ Start (resume) the specified output connector. The connector resumes with the output the pipeline produces from this point on; output produced while it was paused is not replayed. Refer to the connector documentation for more information: https://docs.feldera.com/connectors/orchestration#output-connectors :param view_name: The name of the view that the connector is attached to. :param connector_name: The name of the connector to start. :raises FelderaAPIError: If the connector is not found, or if the pipeline is not running. """ self.client.start_output_connector(self.name, view_name, connector_name) def input_connector_stats( self, table_name: str, connector_name: str ) -> InputEndpointStatus: """ Get the status of the specified input connector. """ return InputEndpointStatus.from_dict( self.client.input_connector_stats(self.name, table_name, connector_name) ) def output_connector_stats( self, view_name: str, connector_name: str ) -> OutputEndpointStatus: """ Get the status of the specified output connector. """ return OutputEndpointStatus.from_dict( self.client.output_connector_stats(self.name, view_name, connector_name) ) def listen(self, view_name: str) -> OutputHandler: """ Follow the change stream (i.e., the output) of the provided view. Returns an output handle to read the changes. When the pipeline is stopped, the handle is dropped. The handle will only receive changes from the point in time when the listener is created. In order to receive all changes since the pipeline started, you can create the pipeline in the `PAUSED` state using :meth:`start_paused`, attach listeners and unpause the pipeline using :meth:`resume`. To read the changes that one input caused, pass the input's completion token to :meth:`.OutputHandler.wait_for_token` before reading the handle. :param view_name: The name of the view to listen to. """ if self.status() not in [ PipelineStatus.PAUSED, PipelineStatus.RUNNING, PipelineStatus.CONCURRENTBOOTSTRAPPING, PipelineStatus.SYNCHRONIZING, ]: raise RuntimeError("Pipeline must be running or paused to listen to output") handler = OutputHandler(self.client, self.name, view_name) handler.start() return handler def foreach_chunk( self, view_name: str, callback: Callable[[pandas.DataFrame, int], None] ): """ Run the given callback on each chunk of the output of the specified view. The callback will only receive changes from the point in time when the listener is created. In order to receive all changes since the pipeline started, you can create the pipeline in the `PAUSED` state using :meth:`start_paused`, attach listeners and unpause the pipeline using :meth:`resume`. :param view_name: The name of the view. :param callback: The callback to run on each chunk. The callback should take two arguments: - **chunk** -> The chunk as a pandas DataFrame - **seq_no** -> The sequence number. The sequence number is a monotonically increasing integer that starts from 0. Note that the sequence number is unique for each chunk, but not necessarily contiguous. Please note that the callback is run in a separate thread, so it should be thread-safe. Please note that the callback should not block for a long time, as by default, backpressure is enabled and will block the pipeline. .. note:: - The callback must be thread-safe as it will be run in a separate thread. """ if self.status() not in [ PipelineStatus.RUNNING, PipelineStatus.PAUSED, PipelineStatus.CONCURRENTBOOTSTRAPPING, PipelineStatus.SYNCHRONIZING, ]: raise RuntimeError("Pipeline must be running or paused to listen to output") event = Event() handler = CallbackRunner( self.client, self.name, view_name, callback, lambda exception: None, event ) handler.start() event.wait() # Statuses in which `wait_for_completion` waits for the pipeline. _WAIT_FOR_COMPLETION_STATUSES = ( PipelineStatus.RUNNING, PipelineStatus.INITIALIZING, PipelineStatus.PROVISIONING, PipelineStatus.BOOTSTRAPPING, PipelineStatus.CONCURRENTBOOTSTRAPPING, PipelineStatus.SYNCHRONIZING, ) def _is_exactly_once(self) -> bool: """True if the pipeline uses exactly-once fault tolerance, which recovers all of its input after a restart.""" ft = self.runtime_config().__dict__.get("fault_tolerance") or {} return ft.get("model") == "exactly_once" def _not_running_error(self, status: PipelineStatus) -> RuntimeError: """The error for a pipeline that is not in a status to complete.""" msg = f"Pipeline must be running to wait for completion (status: {status.name}" try: details = self.deployment_runtime_status_details() except FelderaAPIError: details = None if details: msg += f", details: {details}" return RuntimeError(msg + ")") def wait_for_completion( self, force_stop: bool = False, timeout_s: float | None = None, unavailable_timeout_s: float = 60.0, ): """ Block until the pipeline has completed processing all input records. This method blocks until (1) all input connectors attached to the pipeline have finished reading their input data sources and issued end-of-input notifications to the pipeline, and (2) all inputs received from these connectors have been fully processed and corresponding outputs have been sent out through the output connectors. This method will block indefinitely if at least one of the input connectors attached to the pipeline is a streaming connector, such as Kafka, that does not issue the end-of-input notification. :param force_stop: If True, the pipeline will be forcibly stopped after completion. False by default. No checkpoints will be made. :param timeout_s: Optional. The maximum time (in seconds) to wait for the pipeline to complete. The default is None, which means wait indefinitely. :param unavailable_timeout_s: The maximum time (in seconds) to wait for a pipeline in the UNAVAILABLE status to become available again. The manager reports UNAVAILABLE when one status probe fails, for example when the node is overloaded, so this status is usually transient. 60 seconds by default. :raises RuntimeError: If the pipeline is not running, stays UNAVAILABLE for longer than `unavailable_timeout_s`, restarts during the wait without exactly-once fault tolerance, or returns unknown metrics. """ status = self.status() if ( status not in self._WAIT_FOR_COMPLETION_STATUSES and status != PipelineStatus.UNAVAILABLE ): raise self._not_running_error(status) start_time = time.monotonic() unavailable_since: Optional[float] = None first_incarnation: Optional[UUID] = None # Reused by the warning message below; is_complete() would fetch and # discard the same stats, hiding whether the wait is stalled or just # slow (e.g. mid-transaction-commit). latest_stats: Optional[PipelineStatistics] = None wait_for_completion = LongOperationWarning( logger, lambda elapsed: ( f"still waiting for pipeline {self.name} to complete, " f"waited {elapsed:.1f} seconds ({latest_stats.global_metrics.progress_summary()})" ), lambda elapsed: ( f"pipeline {self.name} completed after {elapsed:.1f} seconds" ), ) while True: if timeout_s is not None: elapsed = time.monotonic() - start_time if elapsed > timeout_s: raise TimeoutError( f"timeout ({timeout_s}s) reached while waiting for" f" pipeline '{self.name}' to complete" ) try: latest_stats = self.stats() except ( FelderaAPIError, FelderaTimeoutError, FelderaCommunicationError, ) as err: status = self.status() if status in self._WAIT_FOR_COMPLETION_STATUSES: raise if status != PipelineStatus.UNAVAILABLE: raise self._not_running_error(status) from err now = time.monotonic() if unavailable_since is None: unavailable_since = now elif now - unavailable_since > unavailable_timeout_s: raise self._not_running_error(status) from err time.sleep(1) continue unavailable_since = None # Without exactly-once fault tolerance, a restarted pipeline can # lose input, so its completion does not cover what we wait for. incarnation = latest_stats.global_metrics.incarnation_uuid if first_incarnation is None: first_incarnation = incarnation elif incarnation != first_incarnation: if not self._is_exactly_once(): raise RuntimeError( f"pipeline '{self.name}' restarted while waiting for" f" completion (incarnation {first_incarnation} became" f" {incarnation}) without exactly-once fault" " tolerance; input received since the last" " checkpoint may be lost" ) logger.info( "pipeline %s restarted while waiting for completion" " (incarnation %s became %s); exactly-once fault tolerance" " replays its input", self.name, first_incarnation, incarnation, ) first_incarnation = incarnation pipeline_complete: Optional[bool] = ( latest_stats.global_metrics.pipeline_complete ) if pipeline_complete is None: raise RuntimeError( "received unknown metrics from the pipeline, pipeline_complete is None" ) elif pipeline_complete: wait_for_completion.done() break wait_for_completion.check() time.sleep(1) if force_stop: self.stop(force=True) def is_complete(self) -> bool: """ Check if the pipeline has completed processing all input records. Returns True if (1) all input connectors attached to the pipeline have finished reading their input data sources and issued end-of-input notifications to the pipeline, and (2) all inputs received from these connectors have been fully processed and corresponding outputs have been sent out through the output connectors. """ return self.stats().global_metrics.pipeline_complete def wait_for_idle( self, idle_interval_s: float = 5.0, timeout_s: float | None = None, poll_interval_s: float = 0.2, ): """ Wait for the pipeline to become idle and then returns. Idle is defined as a sufficiently long interval in which the number of input and processed records reported by the pipeline do not change, and they equal each other (thus, all input records present at the pipeline have been processed). :param idle_interval_s: Idle interval duration (default is 5.0 seconds). :param timeout_s: Timeout waiting for idle (`None` = no timeout is enforced). :param poll_interval_s: Polling interval, should be set substantially smaller than the idle interval (default is 0.2 seconds). :raises ValueError: If idle interval is larger than timeout, poll interval is larger than timeout, or poll interval is larger than idle interval. :raises RuntimeError: If the metrics are missing or the timeout was reached. """ if timeout_s is not None and idle_interval_s > timeout_s: raise ValueError( f"idle interval ({idle_interval_s}s) cannot be larger than" f" timeout ({timeout_s}s)" ) if timeout_s is not None and poll_interval_s > timeout_s: raise ValueError( f"poll interval ({poll_interval_s}s) cannot be larger than" f" timeout ({timeout_s}s)" ) if poll_interval_s > idle_interval_s: raise ValueError( f"poll interval ({poll_interval_s}s) cannot be larger " f"than idle interval ({idle_interval_s}s)" ) start_time_s = time.monotonic() idle_started_s = None prev = (0, 0) while True: now_s = time.monotonic() # Metrics retrieval metrics = self.stats().global_metrics total_input_records = metrics.total_input_records total_processed_records = metrics.total_processed_records if metrics.total_input_records is None: raise RuntimeError( "total_input_records is missing from the pipeline metrics" ) if metrics.total_processed_records is None: raise RuntimeError( """total_processed_records is missing from the pipeline \ metrics""" ) # Idle check unchanged = ( prev[0] == total_input_records and prev[1] == total_processed_records ) equal = total_input_records == total_processed_records prev = (total_input_records, total_processed_records) if unchanged and equal: if idle_started_s is None: idle_started_s = now_s else: idle_started_s = None if idle_started_s is not None and now_s - idle_started_s >= idle_interval_s: return # Timeout if timeout_s is not None and now_s - start_time_s >= timeout_s: raise RuntimeError(f"waiting for idle reached timeout ({timeout_s}s)") time.sleep(poll_interval_s) def _wait_for_adhoc_write( self, poll_interval_s: float = 0.1, timeout_s: float | None = None ): """ Block until every record currently in the circuit has been processed to completion. This backs the ``wait`` option of :meth:`.query` and :meth:`.execute`. An ad-hoc ``INSERT`` returns once its records have been pushed to the circuit, but not necessarily once the circuit has processed them, so a subsequent ad-hoc ``SELECT`` is not guaranteed to observe the write. To close that gap, this method snapshots the number of records that have entered the circuit (``total_input_records - buffered_input_records``) and polls until the pipeline reports at least that many records completed. The write's records are part of that count, because the ``/query`` endpoint pushes them to the circuit before returning. The metric polled is ``total_completed_records``, so "completed" means more than just visible to ad-hoc queries: a record is counted only once the DBSP engine has processed it *and* every output connector has processed the outputs derived from it. Therefore, when this method returns, both of the following hold for the write: - ad-hoc queries reflect the updates, and - all output connectors have processed (sent out) the outputs the updates produced. :param poll_interval_s: Interval between ``/stats`` polls. :param timeout_s: Maximum time to wait, or ``None`` to wait indefinitely. :raises TimeoutError: If ``timeout_s`` elapses before the records are processed. """ metrics = self.stats().global_metrics target = metrics.total_input_records - metrics.buffered_input_records start_s = time.monotonic() while self.stats().global_metrics.total_completed_records < target: if timeout_s is not None and time.monotonic() - start_s >= timeout_s: raise TimeoutError( f"timeout ({timeout_s}s) reached while waiting for an ad-hoc" f" write on pipeline '{self.name}' to be processed" ) time.sleep(poll_interval_s) def activate( self, wait: bool = True, timeout_s: Optional[float] = None ) -> Optional[PipelineStatus]: """ Activates the pipeline when starting from STANDBY mode. Only applicable when the pipeline is starting from a checkpoint in object store. :param wait: Set True to wait for the pipeline to activate. True by default :param timeout_s: The maximum time (in seconds) to wait for the pipeline to pause. """ return self.client.activate_pipeline(self.name, wait=wait, timeout_s=timeout_s) def start( self, bootstrap_policy: Optional[BootstrapPolicy] = None, silent_bootstrap: bool = False, concurrent_bootstrap: bool = False, wait: bool = True, timeout_s: Optional[float] = None, dismiss_error: bool = True, reloadable_tables: Optional[List[str]] = None, ): """ .. _start: Starts this pipeline. - The pipeline must be in STOPPED state to start. - If the pipeline is in any other state, an error will be raised. - If the pipeline is in PAUSED state, use `.meth:resume` instead. :param bootstrap_policy: The bootstrap policy to use. :param silent_bootstrap: Set True to bootstrap the pipeline with output connectors disabled. False by default. :param concurrent_bootstrap: Set True to bootstrap new and modified views concurrently, keeping the pre-existing views live while the new ones backfill in the background. Mutually exclusive with `silent_bootstrap`. False by default. :param timeout_s: The maximum time (in seconds) to wait for the pipeline to start. :param wait: Set True to wait for the pipeline to start. True by default :param dismiss_error: Set True to dismiss any deployment error before starting; set False to make it fail in that case. True by default. :param reloadable_tables: Tables that bootstrapping may reload. Use it with `bootstrap_policy=BootstrapPolicy.ALLOW` and `wait=True`. If a table is modified, bootstrapping will reload the table: it truncates the table and resets its input connectors. Write each name exactly as the pipeline reports it in the approval diff: an unquoted SQL name in lowercase, a quoted name without its quotes. For example, write `t1` for `CREATE TABLE T1` and `MyTable` for `CREATE TABLE "MyTable"`. None by default: ALLOW bootstraps without a check. If no tables are acceptable to reload, set `reloadable_tables=[]`. :raises RuntimeError: If the pipeline is not in STOPPED state. :raises TableReloadNotAllowedError: If a modified table is not in `reloadable_tables`, or the pipeline cannot compute the program diff. The pipeline stays in the AWAITINGAPPROVAL state: call `approve()` to bootstrap or `stop(force=True)` to reject. :raises ValueError: If `reloadable_tables` is set with a bootstrap policy other than ALLOW, or with `wait=False`. """ self.client.start_pipeline( self.name, bootstrap_policy=bootstrap_policy, silent_bootstrap=silent_bootstrap, concurrent_bootstrap=concurrent_bootstrap, wait=wait, timeout_s=timeout_s, dismiss_error=dismiss_error, reloadable_tables=reloadable_tables, ) def start_paused( self, bootstrap_policy: Optional[BootstrapPolicy] = None, silent_bootstrap: bool = False, concurrent_bootstrap: bool = False, wait: bool = True, timeout_s: Optional[float] = None, dismiss_error: bool = True, reloadable_tables: Optional[List[str]] = None, ): """ Starts the pipeline in the paused state. :param bootstrap_policy: The bootstrap policy to use. :param silent_bootstrap: Set True to bootstrap the pipeline with output connectors disabled. False by default. :param concurrent_bootstrap: Set True to bootstrap new and modified views concurrently, keeping the pre-existing views live while the new ones backfill in the background. Mutually exclusive with `silent_bootstrap`. False by default. :param wait: Set True to wait for the pipeline to start. True by default. :param timeout_s: The maximum time (in seconds) to wait for the pipeline to start (defaults to `None` = no timeout is enforced). :param dismiss_error: Set True to dismiss any deployment error before starting; set False to make it fail in that case. True by default. :param reloadable_tables: Tables that bootstrapping may reload. See :meth:`start`. """ return self.client.start_pipeline_as_paused( self.name, bootstrap_policy=bootstrap_policy, silent_bootstrap=silent_bootstrap, concurrent_bootstrap=concurrent_bootstrap, wait=wait, timeout_s=timeout_s, dismiss_error=dismiss_error, reloadable_tables=reloadable_tables, ) def start_standby( self, bootstrap_policy: Optional[BootstrapPolicy] = None, silent_bootstrap: bool = False, concurrent_bootstrap: bool = False, wait: bool = True, timeout_s: Optional[float] = None, dismiss_error: bool = True, reloadable_tables: Optional[List[str]] = None, ): """ Starts the pipeline in the standby state. :param bootstrap_policy: The bootstrap policy to use. :param silent_bootstrap: Set True to bootstrap the pipeline with output connectors disabled. False by default. :param concurrent_bootstrap: Set True to bootstrap new and modified views concurrently, keeping the pre-existing views live while the new ones backfill in the background. Mutually exclusive with `silent_bootstrap`. False by default. :param wait: Set True to wait for the pipeline to start. True by default. :param timeout_s: The maximum time (in seconds) to wait for the pipeline to start (defaults to `None` = no timeout is enforced). :param dismiss_error: Set True to dismiss any deployment error before starting; set False to make it fail in that case. True by default. :param reloadable_tables: Tables that bootstrapping may reload. See :meth:`start`. """ self.client.start_pipeline_as_standby( self.name, bootstrap_policy=bootstrap_policy, silent_bootstrap=silent_bootstrap, concurrent_bootstrap=concurrent_bootstrap, wait=wait, timeout_s=timeout_s, dismiss_error=dismiss_error, reloadable_tables=reloadable_tables, ) def restart( self, bootstrap_policy: Optional[BootstrapPolicy] = None, silent_bootstrap: bool = False, concurrent_bootstrap: bool = False, timeout_s: Optional[float] = None, dismiss_error: bool = True, reloadable_tables: Optional[List[str]] = None, ): """ Restarts the pipeline. This method forcibly **STOPS** the pipeline regardless of its current state and then starts it again. No checkpoints are made when stopping the pipeline. :param bootstrap_policy: The bootstrap policy to use. :param silent_bootstrap: Set True to bootstrap the pipeline with output connectors disabled. False by default. :param concurrent_bootstrap: Set True to bootstrap new and modified views concurrently, keeping the pre-existing views live while the new ones backfill in the background. Mutually exclusive with `silent_bootstrap`. False by default. :param timeout_s: The maximum time (in seconds) to wait for the pipeline to restart. :param dismiss_error: Set True to dismiss any deployment error before starting; set False to make it fail in that case. True by default. :param reloadable_tables: Tables that bootstrapping may reload. See :meth:`start`. """ self.stop(force=True, timeout_s=timeout_s) self.start( bootstrap_policy=bootstrap_policy, silent_bootstrap=silent_bootstrap, concurrent_bootstrap=concurrent_bootstrap, timeout_s=timeout_s, dismiss_error=dismiss_error, reloadable_tables=reloadable_tables, ) def pause(self, wait: bool = True, timeout_s: Optional[float] = None): """ Pause the pipeline. The pipeline can only transition to the PAUSED state from the RUNNING state. If the pipeline is already paused, it will remain in the PAUSED state. :param wait: Set True to wait for the pipeline to pause. True by default :param timeout_s: The maximum time (in seconds) to wait for the pipeline to pause. """ self.client.pause_pipeline(self.name, wait=wait, timeout_s=timeout_s) def stop(self, force: bool, wait: bool = True, timeout_s: Optional[float] = None): """ Stops the pipeline. Stops the pipeline regardless of its current state. :param force: Set True to immediately scale compute resources to zero. Set False to automatically checkpoint before stopping. :param wait: Set True to gracefully shutdown listeners and wait for the |
| You are visitor 01402520 — GeezerVison does not store the pages you look up. Best viewed with Netscape Navigator 4.7 or Internet Explorer 6 at 800×600. No cookies required. |