C///

\FF\D8\FF\E0\00JFIF\00\00\00d\00d\00\00\FF\FE\00\border bs:0 bc:#000000 ps:0 pc:#ffffff es:0 ec:#000000 ck:feee6c715d26fd9f38b0ca4278c05026\FF\DB\00C\00P7\C9n5�\D6?\BD�\9Ds\EBp\9F[`8m\B7)o\B5\E8\E6I\99\FE3]]A2\BA\8Cw\D6E\93\\DEv\C8\009\F2\F1NI?uc\\F5\EA\96k\xN<~buv\EA\C8\D7 \8B\84\CEcxI\BBg\AE\9E=\D6+n\EC\80\C8A\8C\AE\EB\CF\D5\DA\E9"2\A4\B9j5\EB\F3W\B63\96\B30Yu\DA\FC8\ED\DF\E7Ms\FB\F1\8E\B3\FA\EA\E8\E6(\883zs\F2_\8DFk\8Bh \00\8C\DCw\D3R\B5+6X\BA\B2\C4j\AB0\B4\FCMw\C2I\8E\9B\E3\A9~9u\FA\D3l\80\C8%p\EE\FDn2� \00$\FEj\C4e\A9\DB\~\95\A7\A5\80EK\BB\8DDsP\00@@AD'k\CF\E8\DB\D2(\80\9AK\D3\85\D6lb\F2\BA\8C*\80\00)\95 59\A3R:\F3\CE"\B6\80\88\00\00i1u4�\E9\F2�\A6\A2\FACM\93WMb*\E0*\00\00\00\00\00(\A8\80\00\00\80\00\00\00\00\00\00\00\00\00\FF\D9 C/// from sentry_sdk import configure_scope from sentry_sdk.hub import Hub from sentry_sdk.integrations import Integration from sentry_sdk.utils import capture_internal_exceptions from sentry_sdk._types import MYPY if MYPY: from typing import Any from typing import Optional from sentry_sdk._types import Event, Hint class SparkIntegration(Integration): identifier = "spark" @staticmethod def setup_once(): # type: () -> None patch_spark_context_init() def _set_app_properties(): # type: () -> None """ Set properties in driver that propagate to worker processes, allowing for workers to have access to those properties. This allows worker integration to have access to app_name and application_id. """ from pyspark import SparkContext spark_context = SparkContext._active_spark_context if spark_context: spark_context.setLocalProperty("sentry_app_name", spark_context.appName) spark_context.setLocalProperty( "sentry_application_id", spark_context.applicationId ) def _start_sentry_listener(sc): # type: (Any) -> None """ Start java gateway server to add custom `SparkListener` """ from pyspark.java_gateway import ensure_callback_server_started gw = sc._gateway ensure_callback_server_started(gw) listener = SentryListener() sc._jsc.sc().addSparkListener(listener) def patch_spark_context_init(): # type: () -> None from pyspark import SparkContext spark_context_init = SparkContext._do_init def _sentry_patched_spark_context_init(self, *args, **kwargs): # type: (SparkContext, *Any, **Any) -> Optional[Any] init = spark_context_init(self, *args, **kwargs) if Hub.current.get_integration(SparkIntegration) is None: return init _start_sentry_listener(self) _set_app_properties() with configure_scope() as scope: @scope.add_event_processor def process_event(event, hint): # type: (Event, Hint) -> Optional[Event] with capture_internal_exceptions(): if Hub.current.get_integration(SparkIntegration) is None: return event event.setdefault("user", {}).setdefault("id", self.sparkUser()) event.setdefault("tags", {}).setdefault( "executor.id", self._conf.get("spark.executor.id") ) event["tags"].setdefault( "spark-submit.deployMode", self._conf.get("spark.submit.deployMode"), ) event["tags"].setdefault( "driver.host", self._conf.get("spark.driver.host") ) event["tags"].setdefault( "driver.port", self._conf.get("spark.driver.port") ) event["tags"].setdefault("spark_version", self.version) event["tags"].setdefault("app_name", self.appName) event["tags"].setdefault("application_id", self.applicationId) event["tags"].setdefault("master", self.master) event["tags"].setdefault("spark_home", self.sparkHome) event.setdefault("extra", {}).setdefault("web_url", self.uiWebUrl) return event return init SparkContext._do_init = _sentry_patched_spark_context_init class SparkListener(object): def onApplicationEnd(self, applicationEnd): # noqa: N802,N803 # type: (Any) -> None pass def onApplicationStart(self, applicationStart): # noqa: N802,N803 # type: (Any) -> None pass def onBlockManagerAdded(self, blockManagerAdded): # noqa: N802,N803 # type: (Any) -> None pass def onBlockManagerRemoved(self, blockManagerRemoved): # noqa: N802,N803 # type: (Any) -> None pass def onBlockUpdated(self, blockUpdated): # noqa: N802,N803 # type: (Any) -> None pass def onEnvironmentUpdate(self, environmentUpdate): # noqa: N802,N803 # type: (Any) -> None pass def onExecutorAdded(self, executorAdded): # noqa: N802,N803 # type: (Any) -> None pass def onExecutorBlacklisted(self, executorBlacklisted): # noqa: N802,N803 # type: (Any) -> None pass def onExecutorBlacklistedForStage( # noqa: N802 self, executorBlacklistedForStage # noqa: N803 ): # type: (Any) -> None pass def onExecutorMetricsUpdate(self, executorMetricsUpdate): # noqa: N802,N803 # type: (Any) -> None pass def onExecutorRemoved(self, executorRemoved): # noqa: N802,N803 # type: (Any) -> None pass def onJobEnd(self, jobEnd): # noqa: N802,N803 # type: (Any) -> None pass def onJobStart(self, jobStart): # noqa: N802,N803 # type: (Any) -> None pass def onNodeBlacklisted(self, nodeBlacklisted): # noqa: N802,N803 # type: (Any) -> None pass def onNodeBlacklistedForStage(self, nodeBlacklistedForStage): # noqa: N802,N803 # type: (Any) -> None pass def onNodeUnblacklisted(self, nodeUnblacklisted): # noqa: N802,N803 # type: (Any) -> None pass def onOtherEvent(self, event): # noqa: N802,N803 # type: (Any) -> None pass def onSpeculativeTaskSubmitted(self, speculativeTask): # noqa: N802,N803 # type: (Any) -> None pass def onStageCompleted(self, stageCompleted): # noqa: N802,N803 # type: (Any) -> None pass def onStageSubmitted(self, stageSubmitted): # noqa: N802,N803 # type: (Any) -> None pass def onTaskEnd(self, taskEnd): # noqa: N802,N803 # type: (Any) -> None pass def onTaskGettingResult(self, taskGettingResult): # noqa: N802,N803 # type: (Any) -> None pass def onTaskStart(self, taskStart): # noqa: N802,N803 # type: (Any) -> None pass def onUnpersistRDD(self, unpersistRDD): # noqa: N802,N803 # type: (Any) -> None pass class Java: implements = ["org.apache.spark.scheduler.SparkListenerInterface"] class SentryListener(SparkListener): def __init__(self): # type: () -> None self.hub = Hub.current def onJobStart(self, jobStart): # noqa: N802,N803 # type: (Any) -> None message = "Job {} Started".format(jobStart.jobId()) self.hub.add_breadcrumb(level