[ 
https://issues.apache.org/jira/browse/AIRFLOW-1536?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Timothy O'Keefe updated AIRFLOW-1536:
-------------------------------------
    Description: 
All DaemonContext instances used for worker, scheduler, webserver, flower, etc. 
do not supply a umask argument. See here for example:

https://github.com/apache/incubator-airflow/blob/b0669b532a7be9aa34a4390951deaa25897c62e6/airflow/bin/cli.py#L869

As a result, the DaemonContext will use the default umask=0 which leaves user 
data exposed. A BashOperator for example that writes any files would have 
permissions rw-rw-rw- as would any airflow logs.

I believe the umask should be either configurable, or inherited from the parent 
shell, or both.

  was:
All DaemonContext instances used for worker, scheduler, webserver, flower, etc. 
do not supply a umask argument. See here for example:

https://github.com/apache/incubator-airflow/blob/b0669b532a7be9aa34a4390951deaa25897c62e6/airflow/bin/cli.py#L869

As a result, the DaemonContext will use the default umask=0 which leaves user 
data exposed. A BashOperator for example that writes any files would have 
permissions rw-rw-rw- as would any airflow logs.

I believe the umask should be inherited from the parent shell.


> DaemonContext uses default umask 0
> ----------------------------------
>
>                 Key: AIRFLOW-1536
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-1536
>             Project: Apache Airflow
>          Issue Type: Bug
>          Components: cli, security
>            Reporter: Timothy O'Keefe
>
> All DaemonContext instances used for worker, scheduler, webserver, flower, 
> etc. do not supply a umask argument. See here for example:
> https://github.com/apache/incubator-airflow/blob/b0669b532a7be9aa34a4390951deaa25897c62e6/airflow/bin/cli.py#L869
> As a result, the DaemonContext will use the default umask=0 which leaves user 
> data exposed. A BashOperator for example that writes any files would have 
> permissions rw-rw-rw- as would any airflow logs.
> I believe the umask should be either configurable, or inherited from the 
> parent shell, or both.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to