- #1200944564541400: The final state of sub-pipeline tasks that have failed child tasks is now called "Failed" instead of "Upstream Failed".
- #1200875727993229: Observe metric trends over time: missing charts, run state charts, state metrics charts, schema change metric trends, etc.
- #1201110693029241: Operations merged from the SDK and BigQuery logged with
log_dataset_op
are now recognized and displayed accordingly (with BigQuery ID) in the Dataset Operations tab, including with stateful issues. - #1201314083672982: Resolved notifications are no longer the default configuration for Slack receivers.
- #1201141157594482: Alert dialog loads promptly and displays the correct state.
- #1201257264431636: Added support for tracking specific DAGs.
- #1201226401826259: Syncer memory guard limit is now configurable.
- #1201266020003308: Pipelines with zero associated runs are hidden in the Databand UI.
- #1201260807813998: Resolved notifications are now correctly delivered to the update/validate receivers.
- #1201036807083630: In the Alert Definitions' editor, individual tasks can be addressed when DAG has multiple tasks with the same name.
- #1200991140831460:
COPY INTO
queries are now supported for Snowflake. - #1201154986480578: Operations section on the Datasets tab notifies of unusual activity in an operation (schema change or failed operation).
- #1201080247459590: Operations are now tracked with BigQuery even when no SDK is installed.
- #1201080247459582: Dataset Overview section combines the data from SDK and from BigQuery.
- #1201160985639176: Solved a bug that emerged when deleting a DAG.
- #1200764884086970: Added SDK Client for Metadata (basic access API endpoint for collecting all errors in a Run).
- #1201013784336315: Improved behavior of Reset Selection in Affected Datasets.
- #1200766175220432: Now you can create alerts directly from the redesigned Metrics tab.
- #1200964944935744: Each step on the X-axis in Dataset Overview now equals 24 hours from the previous point.
- #1200964944935725: "Reported from" and "Dataset Type" columns in the Datasets tab follow the same logic as elsewhere in the Databand UI.
- #1200638623009192: Datasets can be archived now.
- #1200713291030116: Added empty state clickable component.
- #1200920981691825: Default operation is pre-selected on the Alert definition screen.
- #1200948790296167: Dropped all
python 2.7
support. - #1200766175220434: All alerted metrics that are being reported are now visible.
- #1200892336582865: Reviewed documentation links in the code.
- #1200876485130367: Drop py27 support for dbnd-examples.
- #1200854192669862: Drop support for tensorflow <2.4 and py27 in dbnd-tensorflow plugin.
- #1200463416979272: Run metrics in the list of metrics are now selected by default and are available from the filters in the Databand UI.
- #1200794953399592: A new look for the Dataset page.
- #1200667723456248: Metrics page added - see all the metrics, alerts, and trends for numeric values in one place (coming in a later minor release).
- #1200876485130367: Dropped py27 support for dbnd-examples.
- #1200854192669862: Dropped support for tensorflow <2.4 and py27 in dbnd-tensorflow plugin.
- #1200495012906365: Defined and implemented Network Policy to isolate workloads in namespaces.
- #1200845451577692: Delete alert definitions in bulk using the “Delete” button.
- #1200791698233717: Trigger value of Missing Operation alert now reflects the total number of missing operations in the run.
- #1200667315624593: Tooltip in schema change now opens on click.
- #1200587432732541: Text in the UI now aligns to the left after a line break.
- #1200785050007987: Fixed a bug where the “last modified” field returned an empty value.
- #1200847984199269: Fixed a “type change” reporting bug in schema change alerts.
- #1200526324086051: BigQuery connector - Databand can now sync with your BigQuery tables.
- #1200667621364917: Databand Dashboard can now be shared and will include filters, metrics, and aggregations.
- #1200700177752978: Airflow Monitor V1 is deprecated.
- #1200618993918094: Improved visibility of missing operations in Affected Datasets.
- #1200721895327001: Updated the UI on the Integrations page.
- #1200624166545000: Full table path in the Affected Datasets tab is now a tooltip.
- #1200654355859365: Source and project names are displayed correctly.
- #1200670021315930: Missing operations are now properly represented in the History Trend chart in the Affected Datasets tab.
- #1200655482328132: Fixed the dropdown table column in the 'Add new alert' page.
- #1200619480835694: Pipeline and run lists optimized for tracking.
- #1200597628980930: Now you can choose whether to have alerts fire on historical runs by editing alert definitions.
- #1200347108016370: Range alerts added. Now it’s possible to add an alert if the metric's value is in a predefined range or out of range.
- #1200700419038963: New schema Alert flow changes in Affected Datasets: schema changes in a data operation are now shown in the Affected Datasets screen. Now you'll know when a column is added, removed, or when its type changes. Some minor cosmetic changes in the UI were also implemented.
- #1200610543227298: Simplified usage of
in_cluster
in cases wherein_cluster
is None.
- #1200629604891227: Run view performance improvements.
- #1200535318342009: Databand Alerting now supports email as a notification channel.
- #1200401288822416: Per Project and Source Alert routing - you can define Alerts on any job as you like.
- #1200443526055542: New Run duration metrics: Duration, Total Duration, Compute Time.
- #1200596368463950: New alert type: get notifications when a schema changes on the data you logged with
log_dataset_op
.
- #1200535433472848: Dashboard widgets now show the project or source name if there are multiple sources. This addresses the case of a pipeline with the same name coming from 2 sources.
- #1200587432732540: Look at the affected datasets of past Runs by clicking on Runs.
- #1200557294252243: New tooltip added to the Dataset overview chart - you now can display the Run status and add/view label.
- #1200602972144983: Dataset overview chart changed from stacked to a regular graph.
- #1200526324086040: New alert type: when using
log_dataset_op
, you can get notifications when a pipeline didn’t read or write as expected. - #1200430856087576: Alerts can now be set directly from the Alerts page.
- #1200557294252244: Missing operations are now prominently displayed in the history chart and in the Trend operation bars graph.
- #1200557294252253: Improved coloring for better readability of dataset graphs.
- #1200587833068258: The description in the alert pop-up is no longer missing.
- #1200587301890745: Datasets page sorting and filtering are fixed - dataset operations are sorted by most recent first.
- #1200557970734951: Current Run is marked as usual.
- #1200579848765971: Datasets page filters project correctly.
- #1200404227512965: Choose and view any suitable metric in a new chart in the Favorite Metric widget.
- #1200369882031623: View Run's data operations (writes, schema, etc.) in the Run with alerts and see what happened in the last 6 Runs.
- #1200526450615526: Helm chart variables now support Global image configs.
- #1200526450615534: Improved performance of Databand Web UI configuration. AS-1200371843828304 Errors from Airflow are now displayed in the Run list.
- #1200463339266254: An extensive round of improvements of the metrics reporting and display processes.
- #1200258862047040: Access to custom metrics is now available. Reporting metrics and discovering what metrics are reported within Databand got much easier.
- #1199332360687562: [Breaking change!] Deprecated
__execute
tasks and separation for root-Run and sub-Runs per Airflow task. All the information which was previously reported on<task_name>__execute task
(like metrics or log) is now reported on<task_name> task
. - #1200483852764557: The
log_dataset_op
behavior is now the same forread
andwrite
. - #1200445773961847:
log_dataset_op
can have thelog_df
functionality.
- #1200445773961849: Introducing
log_dataset_op
- now you can manually log the dataset you are interacting with. - #1200130408884030: Outside-of-the-box tracking support for PySpark with Databricks.
- #1199370081934662: Added Live Logs for Kubernetes workers in the Airflow UI.
- #1200401140575119: Missing Data now triggers an alert.
- #1200454655158822: Now you can calculate metric aggregations like SUM/AVG and others and display them in the Favorite charts widget. You can also compare trends over the filtered time period and display trends.
- #1200430623687636: Dataset Screen is deprecated and replaced by Data Targets and Datasets screen in the UI.
- #1200466625626565: Added support for custom headers for Alert Manager webhooks.
- #1200403795815183: Improved Favorite Metric widget usability.
- #1200463339266254: Improved Metric tab interface.
- #1200452512156534: ‘Triggered by column’ is removed from the Run list.
- #1200408168252754: History tab is removed from the Run details.
- #1200430599620006: Timeline Tab in the Run details is deprecated.
- #1200181993522176: Proper support for DagRun submit args and XCom; values are reported with correct types.
- #1200275295497774: Added support for the Airflow retries and restarts in Alerts.
- #1200372371285197: It’s now possible to expose Run env and user in the
stats_api
for alerting.
- #1200402990123287: Fixed a bug that prevented proper tracking from Databricks.
- #1200358546721956: Metrics that cannot be presented in a graph now cannot be added.
- #1200275951379648: Fixed incorrect handling of attempts.
- #1200440061411916: Fixed tracking tasks from
@task
. - #1200035644894886: Fixed browser ‘Back’ button behavior in the Run details page.
- #1200503204031963: Fixed bug that prevented to define advanced alert
- #1200542429205442: Fixed support for airflow_local_settings in auto-tracking
- #1200378732521423: Airflow Monitor v2 now supports Airflow 2.0.
- #1200392175180332: The Favorite Metric widget in the Dashboard now automatically displays anomalies for metrics.
- #1200213466288282: Now supporting Google Composer for Airflow tracking.
- #1200340430085787: Main menu now expands to include titles.
- #1200321463111129: Added a chart display toggle for switching between Run charts in the Favorite Metric widget.
- #1200095672805698: Users can select the main page view. The default mode is Dashboard; options: Dashboard or Pipelines.
- #1200309661632765: The Syncer page was moved to the Settings page.
- #1200284286454068: Now you can identify errors in different Runs - both the error message and the error type now lead to the Run page; the task that triggered the error is selected so you can view the error.
- #1200403425760171: Added new Syncer issue counters in the Settings section.
- #1200420269065898: Only the table is now scrollable in the Metrics widget.
- #1200400792119627: Changes in the chart type design.
- #1200400792119629: Changes in the Dashboard chart tooltips - now, the nearest point tooltip will be displayed.
- #1200251794221201: Improved the readability of the metric charts tooltip.
- #1200378661150501: Added Error Column to All Runs and Pipeline Runs lists.
- #1200369692119674: Improved usability of the Alert Definition section.
- #1200374990956464: Tasks in Airflow sub-processes are now automatically assigned unique names.
- #1200102937873725: Added a unified context manager for handling log dataset operations.
- #1200279378863216: Aligned and improved alert counters in the Pipeline and Run lists.
- #1200401407223048: Fixed custom date picker behavior and placed it next to the Pipeline filter.
- #1200126969897499: Task ID is now correctly displayed in the UI.
- #1200253837838312: Added Deequ + PySpark integration.
- #1200245988366766: Added multiple metrics support in the Favorite Metric widget.
- #1200095672805698: Home Screen selection option is now available in the Settings Page.
- #1200306047670389: It’s now possible to enable the debug mode in PyDev from the CLI.
- #1200179220565213: Slack notifications can now be received from the Databand UI.
- #1199965369989833: Tracking is now compatible with Airflow 2.0.
- #1200213466288282: Airflow tracking is running on Google Composer.
- #1200105822969991: Added support for liveness probe check in Airflow Monitor v2.0.
- #1200327639593933: Latest Pandas version (1.2.x) is now supported.
- #1200223284217045: Improved Alert screen.
- #1200345876827058: Alert rules can be enabled based on the Run env.
- #1200269646547649: “Last Active” widget moved to the bottom of the screen.
- #1200269646547644: Pipeline screen's default view is set to “List”.
- #1200310831229608: PyWheel build tasks are not displayed in the UI now.
- #1200315872631815: Changed the cursor in Favorite Charts.
- #1200305391366264: Fixed missing filters and Run names in the Alert API.
- #1200324498682162: Identified and fixed the issue with Sentry integration.
- #1200275951379648: Fixed incorrect handling of attempts.
- #1200290917819476: Changed Error Widget headline to “Top Errors”.
- #1200359124635729: Fixed error on wrong outputs’ schema in Orchestration.
- #1200183306770565: Stats widget changed to display the top failure rate Pipelines.
- #1200191541441378: Server-side sorting for Pipeline stats widget.
- #1200258224502988: New Anomaly Detection alert is triggered if Custom Metric grows too fast.
- #1200259328525645: Airflow Monitor v.2.
- #1200183150156169: It’s now possible to see an alert’s effect on the affected data.
- #1200044288700180: Drill down to Runs & Pipelines from the Dashboard.
- #1200259770331112: Click iteration and reporting stats added to Dashboard.
- #1200058439151159: Airflow monitor heartbeat indication.
- #1200243606518228: Errors Discovery Flow added for non-Python Operators.
- #1200276060368190: Negative values are now displayed in Dashboard charts.
- #1200285224722668: Fixed Docker version incompatibility issue (extended the number of supported versions).
- #1200275951379659: Fixed Kubernetes execution bug when tasks were wrongly marked as canceled on the end of Pipelines.
- #1168991558489592: Auto-log paths became available for Spark JVM I/O (Alpha version).
- #1200183406021457: It’s now possible to trigger a numeric alert when an anomaly is detected.
- #1200182230484446: Added alerts integration with Opsgenie.
- #1199534439165430: Tracking system error handling is out - enjoy safe tracking without the performance penalty.
- #1200183406021460: See how a newly defined alert would have affected your past runs.
- #1200183406021459: Define sensitivity and the number of runs for anomaly detection.
- #1200194910187394: Slack alert headline now leads to the Alert page and displays the number of firing alerts.
- #1200141500926246: Fixed the URL in Slack message alerts.
- #1200194910187389: Slack alert’s link is now leading to the first alert.
- #1199905812865631: Added Favorite Metrics widget.
- #1199905812865626: Added reference date and fix time range widget.
- #1200036364804040: Added support for Date picker in the UI.
- #1199908347034091: Added Python Spark Metrics listener (similar to JVM).
- #1200135620545740: New alerting mechanism for on-prem users (available on demand).
- #1200130408884041: Zero-cost integration of SQL Airflow DAGs with Snowflake (in DBND SDK).
- #1199510429341979: Configured POC alerting.
- #1199592642123487: Run name/run description is now editable from the UI.
- #1199917986044490: It’s now possible to access the latest webserver logs via the webserver.
- #1200092086191209: Added a descriptive error message that appears when installing Airflow Monitor not in the same folder as Airflow.
- #1200030168701007: Support for the cloud UI is here. Now you can gain access to the Databand UI in the cloud.
- #1200016204502957: Users who have Airflow with Docker Compose environment can now use the Databand cloud solution.
- #1200058439151184: A flag for switching on\off the transmission (sending) of the source code was added.
- #1199921442266004: AlertDef history now displays seconds in the UI.
- #1199995565427900: Version and Environment badges are now displayed on the Login page.
- #1200093251774759: Graphic info card in the Dashboard is now black.
- #1200127804370172: Quick updates tab added to the integration page.
- #1199657019529688:
QueryApi.force_pagination
is now enabled by default. - #1200126969897487: Remote logs are truncated.
- #1200058448826966: Support for "zero computational cost" data insights and preview added.
- #1200129802903561: Databand logo back on the Login page.
- #1200111167427102: Time format bug is fixed.
- #1199965226804188: The search bar is back in Details view.
- #1200091724058557: Runs and Task by Start Time are a stacked graph.
- #1199594880117673: Introducing the Cloud Environment of Databand Web server.
- #1199643165776154: New Design and filters for Run List, including sticky filters.
- #1199914881669553: Project filters now support Airflow tags.
- #1199914881669552: Azkaban projects are now supported.
- #1199371447621207: You can now track Log paths from Spark.
- #1199674986608158: Extended tracking of Operational Health in the Dashboard: see Runs and Task Runs over time.
- #1199965687445132: Added columns for the task table in the Dashboard.
- #1199917986044517: Added Spark LogPath.
- #1199534085669848: Orchestration error Reporting
- #1199382295296733: Project phase 1
- #1199562651279654: Support multiple Azkaban instances
- #1199503238625634: Support user access token
- #1199410590823866: Using pipeline(=band) outputs in runtime
- #1199512537243676: Enable pod logs collection with google logs
- #1199511121032309: Zombie Driver at GKE - unsynced state
- #1199538657255959: Link at Error banner at Details page looks broken
- #1199624982127115: Configuration per task run and not all pipeline run
- #1199560164027872: Zombie Drive at GKE - watcher process not dying on driver termination
- #1199665645625989: fix Marshalling of nested lists
- #1199692629063013: log_dataframe(...,with_stats=True) does not print stats to CLI
- #1199539989855853: silence logging on tracking in production
- #1199410183946263: [java sdk] log collection - bring both head and tail
- #1186815542031177: add @data_source_pipeline annotation
- #1199601902803965: Default delay is set to 1 for log_snowflake_resource_usage
- #1183486068581313: auto-add dbnd spark listener
- #1195941695830497: Clean console outputs in tracking scenario
- #1199534865809603: fix DbndSchedulerDBDagsProvider retries
- #1199553897458772: Pendulum fix for log_metric
- #1199592885591350: EKS(AWS) submitter log read timeout after 4h
- #1199383391833788: Normalize
- #1199513772139741: read azkaban job properties from file
- #1199222341211949: log_metric() is compatible with BIGINTs
- #1199191513095141: remove git working from dbnd log
- #1199352826744440: Better run name and job name for azkaban
- #1199362041298460: add Flow/Job properties to azkaban tracking
- #1198203153122085: introduce airflow uid (sync and execute)
- #1199593868894397: Sync monitor - bad "from"
- #1199562298463958: fix expire_heartbeat query
- #1199561667659647: Move ScheduledJob, RunState updates are done from for loop to bulk
- #1199561667659637: Improve Airflow monitor performance
- #1199360347022221: fix dbnd db init printouts
- #1199351713607668: log collection - bring both head and tail
- #1199222341211949: make log_metric() compatible with BIGINTs
- #1199191513095141: remove git working from log
- #1199352826744440: Better run name and job name for Azkaban
- #1185047487019177: add dataframe name to target reporting with log_dataframe()
- #1198956007770574: fix jvm tracking timeouts
- #1196324185480705: make databand task return proper result
- #1198935333953725: fix pickling compatibility issue for fat wheel.
- #1195948952063399: Azkaban support
- #1198875598805805: Support dbnd_cache for multiple scenarios
- #1199199113928119: Documentation for Snowflake Resources Operator
- #1185230434439866: Mark task that received signal (sigterm?) with different color/state
- #1199191096927422: snowflake: sleep for resource fetch
- #1199109359508985: improve performance of airflow-monitor
- #1199135775563013: add cli to manage airflow monitors
- #1188613755173249: lazy read for task params
- #1199195747557789: improve performance on relation fetching
- #1199183015685020: fix nested run issue
- #1198912139427114: fix airflow logging
- #1198951004489095: task_visualizer banner - configurable max value size
- #1198182294540228: Add Deeque support
- #1196095832101313: Define alerts cli
- #1198215411158397: Standalone tracking and track_operator method
- #1196565777511077: Live logging for all ad-hoc tasks
- #1189443575273138: JVM SDK Logging
- #1189866958488326: Add separate config section for tracking
- #1195079839215316: Support tensorflow 2.0
- #1195941695824501: Support Airflow DAG changes / deletions
- #1198172260918388: Helm - support any env variables block( common for all our services)
- #1125138619443457: Command li Add flag to command line
- #1198199342753981: move require local sync to parameter from target config
- #1194785058398298: track airflow templated parameters
- #1198178723258493: Task-level Airflow link leads to the wrong destination.
- #1179311676368701: Too old resource version on k8s
- #1177515786548674: support MultiTarget with require_local_access=true
- #1194826730548337: Handle incomplete output scenario
- #1195944991344117: add dbnd-snowflake module for tracking snowflake data and metrics
- #1196277825236673: support RBAC for export_plugin
- #1176447682585577: support basic authentication with livy
- #1194347540945401: add rendered xcom to task params
- #1194347540945407: report Airflow DAG params
- #1195924719351296: report Airflow operator params
- #1193537416358473: column type not shown when histograms = true but statistics=False
- #1188126160239994: histograms for pandas on empty column produce an error in a log
- #1191658141334873: Make dbnd-luigi workable
- #1191868676285987: calculate spark descriptive statistics efficiently
- #1192562605942417: Logging in Jupyter Notebok
- #1187220511775489: report target load/save/download time as a metric
- #1192948860705935: fat_build_task inherits target_date from parent
- #1164673115500855: simplifying DBND spark implementation ( no airflow)
- #1193537416358473: column type not shown when histograms = true but statistics=False
- #1188126160239994: histograms for pandas on empty column produce an error in a log
- #1190345801510096: report user params for Airflow operators
###Improvements:
- #1191658141334875: Better exception handling to prevent failing DAG retireval from versioned-dag
- #1188403453583067: HDFS local folder open after pyspark
- #1190536560972406: airflow skipped task has two run attempts in databand db
- #1192515450649121: spark histogram error: list indices must be integers or slices, not NoneType
- #1191864220891825: histogram calculation fails on complex type
- #1190570128931897: Solve compatibility errors of versioned dag and airflow 1.10.10
- #1171185232799219: support --set run.task
- #1184271040807809: Parallelize checking if tasks are complete
- #1189508556532473: support airflow 1.10.10
- #1183936916078880: Retrieve spark exception when running spark locally
- #1186838986890801: Created @log_duration to measure and log execution time as a metric
- #1186434551886169: Log pandas histogram calculation time as a metric
- #1183936916078873: Display JSON path as well as content in preview
- #1185884345508633: Add Livy url to external links
- #1183936916078876: Retrieve Qubole traceback and exception when run fails
- #1187615130069532: ci/cd deploy dbnd to quoble/emr from s3 bucket
- #1184859956507691: airflow monitor: print proper error message on RBAC
- #1187144661730497: Added airflow icon to DAG and Operator levels of the graph
- #1187144661730508: Show recent metric values when selected anomaly detection alert
- #1186620728186105: Link from alert list modal to specific run details page
- #1178184224046201: Tracking is slowing down tasks when dbnd-web address is wrong
- #1188853509921970: Add task-version now to fat_wheel building task
- #1188398439358980: Bring back loading of .dbnd/databand-test.cfg if exist
- #1187836261701377: Fix kubernetes infinite loop on change_state -> delete_pod
- #1188150247653234: support py-files with inline in all engines
- #1189443575273132: Fixed spark and java task graph node icons
- #1186213205378244: Run list table style alignment
- #1187782447252866: Pandas fails to write parquet file to hdfs
- #1184635774415961: Added ability to disable automatic tracking on functions, decorators, and DAGs
- #1186666107312362: Warning message displayed when not all outputs are available for a task.
- #1184186845180735: Support for py_files in the Databricks environment.
- #1185262786942811: Added overwriting of targets via save_options to historically overwrite new data to the same path.
- #1184836053179881: Added ability to configure histograms
- #1186033707483059: Trackers added to a task/pipeline run banner
- #1185174851651484: Tracking logic moved to dbnd_airflow
- #1183947756269387: Changed the default SparkDataFrameToCsv marshaller behaviour
- #1186029239496897: Added automatic logging for Spark initialization
- #1180558848817017: Automatic configuration of the databand_url on Spark
- Timeline - parent tasks now clearly show when a child task was running
- Show value inline for short params and results instead of the path or a placeholder
- #1186251402357473: --disable-web-tracker function does not work.
- #1183600904070465: 500 HTTP error thrown on bad paramater definition: Format=parameter.choices(["artemis","bago_dialer"]).default("artemis")
- #1186666107312349: Recreated filesystem objects on remote execution to prevent credentials conflict
- Timeline - children didn't always appear below their parents
- Timeline - tasks with the same name were incorrectly grouped together
- Refresh button wasn't working correctly on several pages
- Recent values in alert config showing non-recent values
- fixed run list UI issues
- fixed task sometimes not being selected in the graph during initial page load
- large performance improvements for the stats endpoint exporting metrics to prometheus
- Support for fat_wheel for Spark py_files
- New method for canceling pipelines
- Processing tasks all output
- Added ability to move objects across multiple environments
- Logging marshalling operations in DBND
- New pods retry behavior for pipelines running on Kubernetes