You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ambari.apache.org by "Robert Levas (JIRA)" <ji...@apache.org> on 2014/11/29 12:29:12 UTC

[jira] [Updated] (AMBARI-8481) Flume service components should indicate security state

     [ https://issues.apache.org/jira/browse/AMBARI-8481?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Robert Levas updated AMBARI-8481:
---------------------------------
    Description: 
The Flume service components should indicate security state when queried by Ambari Agent via STATUS_COMMAND.  Each component should determine it's state as follows:

h3. FLUME_HANDLER
h4. Indicators
* Command JSON
** config\['configurations']\['cluster-env']\['security_enabled'] 
*** = “true”
* Configuration File: params.flume_conf_dir + _agent?_ + ‘flume.conf’
** agent.sinks.sink-*.hdfs.kerberosKeytab
*** not empty
*** path exists and is readable
*** required
** agent.sinks.sink-1.hdfs.kerberosPrincipal
*** not empty
*** required

h4. Pseudocode
{code}
if indicators imply security is on and validate
    if kinit(flume principal) succeeds
        state = SECURED_KERBEROS
    else
        state = ERROR 
else
    state = UNSECURED
{code}

_*Note*_: Due to the _cost_ of calling {{kinit}} results should be cached for a period of time before retrying.  This may be an issue depending on the frequency of the heartbeat timeout.

_*Note*_: It is possible that multiple sinks may spread out into different _command target_ flume.conf files (in  /etc/flume/conf/<command target dir>/flume.conf)


  was:
The Flume service components should indicate security state when queried by Ambari Agent via STATUS_COMMAND.  Each component should determine it's state as follows:

h3. FLUME_HANDLER
h4. Indicators
* Command JSON
** config\['configurations']\['cluster-env']\['security_enabled'] 
*** = “true”
* Configuration File: /etc/flume/conf/*/flume.conf
** agent.sinks.sink-*.hdfs.kerberosKeytab
*** not empty
*** path exists and is readable
*** required
** agent.sinks.sink-1.hdfs.kerberosPrincipal
*** not empty
*** required

h4. Pseudocode
{code}
if indicators imply security is on and validate
    if kinit(flume principal) succeeds
        state = SECURED_KERBEROS
    else
        state = ERROR 
else
    state = UNSECURED
{code}

_*Note*_: Due to the _cost_ of calling {{kinit}} results should be cached for a period of time before retrying.  This may be an issue depending on the frequency of the heartbeat timeout.

_*Note*_: It is possible that multiple sinks may spread out into different _command target_ flume.conf files (in  /etc/flume/conf/<command target dir>/flume.conf)



> Flume service components should indicate security state
> -------------------------------------------------------
>
>                 Key: AMBARI-8481
>                 URL: https://issues.apache.org/jira/browse/AMBARI-8481
>             Project: Ambari
>          Issue Type: Improvement
>          Components: ambari-server, stacks
>    Affects Versions: 2.0.0
>            Reporter: Robert Levas
>            Assignee: Robert Levas
>              Labels: flume, kerberos, security
>             Fix For: 2.0.0
>
>
> The Flume service components should indicate security state when queried by Ambari Agent via STATUS_COMMAND.  Each component should determine it's state as follows:
> h3. FLUME_HANDLER
> h4. Indicators
> * Command JSON
> ** config\['configurations']\['cluster-env']\['security_enabled'] 
> *** = “true”
> * Configuration File: params.flume_conf_dir + _agent?_ + ‘flume.conf’
> ** agent.sinks.sink-*.hdfs.kerberosKeytab
> *** not empty
> *** path exists and is readable
> *** required
> ** agent.sinks.sink-1.hdfs.kerberosPrincipal
> *** not empty
> *** required
> h4. Pseudocode
> {code}
> if indicators imply security is on and validate
>     if kinit(flume principal) succeeds
>         state = SECURED_KERBEROS
>     else
>         state = ERROR 
> else
>     state = UNSECURED
> {code}
> _*Note*_: Due to the _cost_ of calling {{kinit}} results should be cached for a period of time before retrying.  This may be an issue depending on the frequency of the heartbeat timeout.
> _*Note*_: It is possible that multiple sinks may spread out into different _command target_ flume.conf files (in  /etc/flume/conf/<command target dir>/flume.conf)



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)