[jira] [Closed] (AIRFLOW-4065) misc security fixes and custom spark poll

["Ash Berlin-Taylor (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/AIRFLOW-4065?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Ash Berlin-Taylor closed AIRFLOW-4065.
--------------------------------------
    Resolution: Invalid

This is more than a single issue. Can you open an issue per separate problem that is still valid.

> misc security fixes and custom spark poll
> -----------------------------------------
>
>                 Key: AIRFLOW-4065
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-4065
>             Project: Apache Airflow
>          Issue Type: Bug
>          Components: security
>    Affects Versions: 1.10.2, 1.10.3
>            Reporter: t oo
>            Priority: Major
>
> 1.
> www/app.py Add Click jacking defence
>  
> Fix:
> at the end of     
> def create_app(config=None, testing=False):
> @app.after_request
>  def apply_caching(response):
>  response.headers["X-Frame-Options"] = "DENY"
>  return response
>  
> 2.
> www/app.py Add WebUI login timeout of 15 minutes
>  
> Fix:
> at the end of     
> def create_app(config=None, testing=False):
> @app.before_request
>  def before_request():
>  flask.session.permanent = True
>  app.permanent_session_lifetime = datetime.timedelta(minutes=15)
>  flask.session.modified = True
>  flask.g.user = flask_login.current_user
>  
> 3.
> www/views.py Add Cross Site Scripting defence
> *BEFORE*
> return self.render(
>  'airflow/dags.html',
>  webserver_dags=webserver_dags_filtered,
>  orm_dags=orm_dags,
>  hide_paused=hide_paused,
>  current_page=current_page,
>  search_query=arg_search_query if arg_search_query else '',
>  page_size=dags_per_page,
>  num_of_pages=num_of_pages,
>  num_dag_from=start + 1,
>  num_dag_to=min(end, num_of_all_dags),
>  num_of_all_dags=num_of_all_dags,
>  paging=wwwutils.generate_pages(current_page, num_of_pages,
>  search=arg_search_query,
>  showPaused=not hide_paused),
>  dag_ids_in_page=page_dag_ids,
>  auto_complete_data=auto_complete_data)
>  
> *AFTER*
> return self.render(
>  'airflow/dags.html',
>  webserver_dags=webserver_dags_filtered,
>  orm_dags=orm_dags,
>  hide_paused=hide_paused,
>  current_page=current_page,
>  search_query=arg_search_query if arg_search_query else '',
>  page_size=dags_per_page,
>  num_of_pages=num_of_pages,
>  num_dag_from=start + 1,
>  num_dag_to=min(end, num_of_all_dags),
>  num_of_all_dags=num_of_all_dags,
>  paging=wwwutils.generate_pages(current_page, num_of_pages,
>  search=escape(arg_search_query) if arg_search_query else None,
>  showPaused=not hide_paused),
>  dag_ids_in_page=page_dag_ids,
>  auto_complete_data=auto_complete_data)
>  
> 4.
> contrib/hooks/spark_submit_hook.py Poll spark server at a custom interval instead of every second
>  
> *BEFORE*
>  # Sleep for 1 second as we do not want to spam the cluster
>  time.sleep(1)
>  
> *AFTER*
> import airflow
>  from airflow import configuration as conf
> Sleep for n second as we do not want to spam the cluster
>  _poll_interval = conf.getint('sparksubmit', 'poll_interval')
>  time.sleep(_poll_interval)
>  
> 5. DOCO only. Securing connection to mysql backend metastore.
> At the end of sql_alchemy_conn line in airflow.cfg add    ?ssl_ca=<PEMCERTFORMYSQL.pem>
>  
> 6.
> contrib/hooks/spark_submit_hook.py    
> Mask passwords in spark submit cmd AND error stacktrace
>  
> *add*
> def _mask_cmd(self, connection_cmd):
>  # Mask any password related fields in application args with key value pair where key contains password (case insensitive), e.g. HivePassword='abc'
>  connection_cmd_masked = re.sub(r"(\S*?(?:secret|password)\S*?\s*=\s*')[^']*(?=')", r'\1******', ' '.join(connection_cmd), flags=re.I)
> return connection_cmd_masked
>  
> *BEFORE*
> self.log.info("Spark-Submit cmd: %s", connection_cmd)
>  
> *AFTER*
> self.log.info("Spark-Submit cmd: %s", self._mask_cmd(connection_cmd))
>  
> *BEFORE*
>  if returncode or (self._is_kubernetes and self._spark_exit_code != 0):
>  raise AirflowException(
>  "Cannot execute: {}. Error code is: {}.".format(
>  spark_submit_cmd, returncode
>  )
>  )
>  
> *AFTER*
> if returncode or (self._is_kubernetes and self._spark_exit_code != 0):
>  raise AirflowException(
>  "Cannot execute: {}. Error code is: {}.".format(
>  self._mask_cmd(spark_submit_cmd), returncode
>  )
>  )
>  
> 7. cli.py is currently printing logs with password if you use cli to add connection with conn_password.
> example log is being printed (this is issue if you have a auto-logforwarder like splunk)
> Successfully added `conn_id`=query_hive : hive_cli://user:cleartextpassw@host:10000/default
>  
> relevant code doing the printing:
> with db.create_session() as session:
>  if not (session.query(Connection)
>  .filter(Connection.conn_id == new_conn.conn_id).first()):
>  session.add(new_conn)
>  msg = '\n\tSuccessfully added `conn_id`=\{conn_id} : \{uri}\n'
>  msg = msg.format(conn_id=new_conn.conn_id,
>  uri=args.conn_uri or
>  urlunparse((args.conn_type,
>  '\{login}:\{password}@\{host}:\{port}'
>  .format(login=args.conn_login or '',
>  password=args.conn_password or '',
>  host=args.conn_host or '',
>  port=args.conn_port or ''),
>  args.conn_schema or '', '', '', '')))
>  
> 8. below should be 'airflow.api.auth.backend.deny_all' by default:
> |[api]|
> | |# How to authenticate users of the API|
> | |auth_backend = airflow.api.auth.backend.default|
>  
> 9. defaults in examples should be 1970 to allow historical DAGruns before 2015 as can be confusing
> instead of 'start_date': datetime(2015, 6, 1),
>  



--
This message was sent by Atlassian Jira
(v8.3.4#803005)