You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ambari.apache.org by "Doroszlai, Attila (JIRA)" <ji...@apache.org> on 2018/10/16 17:10:00 UTC

[jira] [Commented] (AMBARI-24774) Wrong settings in exported blueprint

    [ https://issues.apache.org/jira/browse/AMBARI-24774?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16652087#comment-16652087 ] 

Doroszlai, Attila commented on AMBARI-24774:
--------------------------------------------

During cluster creation a service-level {{recovery_enabled: true}} setting in the blueprint results in _all_ components of that service not explicitly listed under {{"component_settings"}} to have "Auto Start" enabled.  Similarly, cluster-level {{recovery_enabled: true}} results in all components having "Auto Start" enabled.

Exported blueprints, however, have service/cluster-level {{recovery_enabled: true}} if _any_ component has "Auto Start" enabled.  This causes further discrepancy between original cluster and re-created cluster.

STR:

# Create ZooKeeper + Metrics cluster without any explicit recovery settings
   result: "Auto Start" is enabled only for Metrics Collector by default
# Export blueprint
# Create cluster using the exported blueprint
   result: "Auto Start" is enabled for every component

> Wrong settings in exported blueprint
> ------------------------------------
>
>                 Key: AMBARI-24774
>                 URL: https://issues.apache.org/jira/browse/AMBARI-24774
>             Project: Ambari
>          Issue Type: Bug
>          Components: ambari-server
>            Reporter: Doroszlai, Attila
>            Assignee: Doroszlai, Attila
>            Priority: Major
>             Fix For: 2.8.0
>
>
> Cluster blueprint export is broken wrt. the {{settings}} section.
> # Only one component per service is included.
> # Default setting value assumed to be {{false}}, ignoring component-level stack definition
> STR:
> # Disable "Auto Start" for Metrics Collector, for which it is enabled by default per stack definition
> # Enable "Auto Start" for both DataNode and NameNode
> # Export blueprint
> Result: 
> {noformat:title=http://$AMBARI_SERVER:8080/api/v1/clusters/TEST?format=blueprint}
>   "settings" : [
>     {
>       "recovery_settings" : [
>         {
>           "recovery_enabled" : "true"
>         }
>       ]
>     },
>     {
>       "service_settings" : [
>         {
>           "recovery_enabled" : "true",
>           "name" : "HDFS"
>         }
>       ]
>     },
>     {
>       "component_settings" : [
>         {
>           "recovery_enabled" : "true",
>           "name" : "NAMENODE"
>         }
>       ]
>     }
> {noformat}
> Problem: creating a cluster using the exported blueprint results in different settings than the original cluster.  "Auto Start" would be disabled for DataNode and enabled for Metrics Collector per stack defaults.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)