-------------------------
- Fix : Don't treat premature tasks as could_not_run tasks
- AIRFLOW-92 Avoid unneeded upstream_failed session closes apache/incubator-airflow1485
- Add logic to lock DB and avoid race condition
- Handle queued tasks from multiple jobs/executors
- AIRFLOW-52 Warn about overwriting tasks in a DAG
- Fix corner case with joining processes/queues (1473)
- [AIRFLOW-52] Fix bottlenecks when working with many tasks
- Add columns to toggle extra detail in the connection list view.
- Log the number of errors when importing DAGs
- Log dagbag metrics dupplicate messages in queue into Statsd (1406)
- Clean up issue template (1419)
- correct missed arg.foreground to arg.daemon in cli
- Reinstate imports for github enterprise auth
- Use os.execvp instead of subprocess.Popen for the webserver
- Revert from using "--foreground" to "--daemon"
- Implement a Cloudant hook
- Add missing args to `airflow clear`
- Fixed a bug in the scheduler: num_runs used where runs intended
- Add multiprocessing support to the scheduler
- Partial fix to make sure next_run_date cannot be None
- Support list/get/set variables in the CLI
- Properly handle BigQuery booleans in BigQuery hook.
- Added the ability to view XCom variables in webserver
- Change DAG.tasks from a list to a dict
- Add support for zipped dags
- Stop creating hook on instantiating of S3 operator
- User subquery in views to find running DAGs
- Prevent DAGs from being reloaded on every scheduler iteration
- Add a missing word to docs
- Document the parameters of `DbApiHook`
- added oracle operator with existing oracle hook
- Add PyOpenSSL to Google cloud gcp_api.
- Remove executor error unit test
- Add DAG inference, deferral, and context manager
- Don't return error when writing files to Google cloud storage.
- Fix GCS logging for gcp_api.
- Ensure attr is in scope for error message
- Fixing misnamed PULL_REQUEST_TEMPLATE
- Extract non_pooled_task_slot_count into a configuration param
- Update plugins.rst for clarity on the example (1309)
- Fix s3 logging issue
- Add twitter feed example dag
- Github ISSUE_TEMPLATE & PR_TEMPLATE cleanup
- Reduce logger verbosity
- Adding a PR Template
- Add Lucid to list of users
- Fix usage of asciiart
- Use session instead of outdated main_session for are_dependencies_met
- Fix celery flower port allocation
- Fix for missing edit actions due to flask-admin upgrade
- Fix typo in comment in prioritize_queued method
- Add HipchatOperator
- Include all example dags in backfill unit test
- Make sure skipped jobs are actually skipped
- Fixing a broken example dag, example_skip_dag.py
- Add consistent and thorough signal handling and logging
- Allow Operators to specify SKIPPED status internally
- Update docstring for executor trap unit test
- Doc: explain the usage of Jinja templating for templated params
- Don't schedule runs before the DAG's start_date
- Fix infinite retries with pools, with test
- Fix handling of deadlocked jobs
- Show only Airflow's deprecation warnings
- Set DAG_FOLDER for unit tests
- Missing comma in setup.py
- Deprecate *args and **kwargs in BaseOperator
- Raise deep scheduler exceptions to force a process restart.
- Change inconsistent example DAG owners
- Fix module path of send_email_smtp in configuration
- added Gentner Lab to list of users
- Increase timeout time for unit test
- Fix reading strings from conf
- CHORE - Remove Trailing Spaces
- Fix SSHExecuteOperator crash when using a custom ssh port
- Add note about airflow components to template
- Rewrite BackfillJob logic for clarity
- Add unit tests
- Fix miscellaneous bugs and clean up code
- Fix logic for determining DagRun states
- Make SchedulerJob not run EVERY queued task
- Improve BackfillJob handling of queued/deadlocked tasks
- Introduce ignore_depends_on_past parameters
- Use Popen with CeleryExecutor
- Rename user table to users to avoid conflict with postgres
- Beware of negative pool slots.
- Add support for calling_format from boto to S3_Hook
- Add pypi meta data and sync version number
- Set dags_are_paused_at_creation's default value to True
- Resurface S3Log class eaten by rebase/push -f
- Add missing session.commit() at end of initdb
- Validate that subdag tasks have pool slots available, and test
- Use urlparse for remote GCS logs, and add unit tests
- Make webserver worker timeout configurable
- Fixed scheduling for once interval
- Use psycopg2's API for serializing postgres cell values
- Make the provide_session decorator more robust
- update link to Lyft's website
- use num_shards instead of partitions to be consistent with batch ingestion
- Add documentation links to README
- Update docs with separate configuration section
- Fix airflow.utils deprecation warning code being Python 3 incompatible
- Extract dbapi cell serialization into its own method
- Set Postgres autocommit as supported only if server version is < 7.4
- Use refactored utils module in unit test imports
- Add changelog for 1.7.0
- Use LocalExecutor on Travis if possible
- remove unused logging,errno, MiniHiveCluster imports
- remove extra import of logging lib
- Fix required gcloud version
- Refactoring utils into smaller submodules
- Properly measure number of task retry attempts
- Add function to get configuration as dict, plus unit tests
- Merge branch 'master' into hivemeta_sasl
- Add wiki link to README.md
- [hotfix] make email.Utils > email.utils for py3
- Add the missing "Date" header to the warning e-mails
- Add the missing "Date" header to the warning e-mails
- Check name of SubDag class instead of class itself
- [hotfix] removing repo_token from .coveralls.yml
- Set the service_name in coverals.yml
- Fixes 1223
- Update Airflow docs for remote logging
- Add unit tests for trapping Executor errors
- Make sure Executors properly trap errors
- Fix HttpOpSensorTest to use fake resquest session
- Linting
- Add an example on pool usage in the documentation
- Add two methods to bigquery hook's base cursor: run_table_upsert, which adds a table or updates an existing table; and run_grant_dataset_view_access, which grants view access to a given dataset for a given table.
- Tasks references upstream and downstream tasks using strings instead of references
- Fix typos in models.py
- Fix broken links in documentation
- [hotfix] fixing the Scheduler CLI to make dag_id optional
- Update link to Common Pitfalls wiki page in README
- Allow disabling periodic committing when inserting rows with DbApiHook
- added Glassdoor to "who uses airflow"
- Fix typo preventing from launching webserver
- Documentation badge
- Fixing ISSUE_TEMPLATE name to include .md suffix
- Adding an ISSUE_TEMPLATE to ensure that issues are adequately defined
- Linting & debugging
- Refactoring the CLI to be data-driven
- Updating the Bug Reporting protocol in the Contributing.md file
- Fixing the docs
- clean up references to old session
- remove session reference
- resolve conflict
- clear xcom data when task instance starts
- replace main_session with provide_session
- Add extras to installation.rst
- Changes to Contributing to reflect more closely the current state of development.
- Modifying README to link to the wiki committer list
- docs: fixes a spelling mistake in default config
- Set killMode to 'control-group' for webservice.service
- Set KillMode to 'control-group' for worker.service
- Linting
- Fix WebHdfsSensor
- Adding more licenses to pass checks
- fixing landscape's config
- [hotfix] typo that made it in master
- [hotfix] fixing landscape requirement detection
- Make testing on hive conditional
- Merge remote-tracking branch 'upstream/master' into minicluster
- Update README.md
- Throwing in a few license to pass the build
- Adding a reqs.txt for landscape.io
- Pointing to a reqs file
- Some linting
- Adding a .landscape.yml file
- badge for pypi version
- Add license and ignore for sql and csv
- Use correct connection id
- Use correct table name
- Provide data for ci tests
- new badge for showing staleness of reqs
- removing requirements.txt as it is uni-dimensional
- Make it work on py3
- Remove decode for logging
- Also keep py2 compatible
- More py3 fixes
- Convert to bytes for py3 compat
- Make sure to be py3 compatible
- Use unicodecsv to make it py3 compatible
- Replace tab with spaces Remove unused import
- Merge remote-tracking branch 'upstream/master'
- Support decimal types in MySQL to GCS
- Make sure to write binary as string can be unicode
- Ignore metastore
- More impyla fixes
- Test HivemetaStore if python 2
- Allow users to set hdfs_namenode_principal in HDFSHook config
- Add tests for Hiveserver2 and fix some issues from impyla
- Merge branch 'impyla' into minicluster
- This patch allows for testing of hive operators and hooks. Sasl is used (NoSasl in connection string is not possible). Tests have been adjusted.
- Treat SKIPPED and SUCCESS the same way when evaluating depends_on_past=True
- fix bigquery hook
- version cap for gcp_api
- Fix typo when returning VerticaHook
- Adding fernet key to use it as part of stdout commands
- Adding support for ssl parameters. (picking up from jthomas123)
- more detail in error message.
- make sure paths don't conflict bc of trailing /
- change gcs_hook to self.hook
- refactor remote log read/write and add GCS support
- Only use multipart upload in S3Hook if file is large enough
- Merge branch 'airbnb/master'
- Add GSSAPI SASL to HiveMetaStoreHook.
- Add warning for deprecated setting
- Use kerberos_service_name = 'hive' as standard instead of 'impala'.
- Use GSSAPI instead of KERBEROS and provide backwards compatibility
- ISSUE-1123 Use impyla instead of pyhs2
- set celery_executor to use queue name as exchange