[ 
https://issues.apache.org/jira/browse/AMBARI-6468?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Aleksandr Kovalenko updated AMBARI-6468:
----------------------------------------

    Description: 
In Ambari-1.6.0 release, we did not created JournalNode and ZKFC service 
components while installing a cluster and created it in NameNode HA wizard.

In Ambari-1.6.1 dev cycle we started creating all service components for all 
selected services including JournalNode and ZKFC. After we removed creating 
JournalNoe and ZKFC in NameNode HA wizard. This creates problem when cluster is 
installed via blueprint and then namennode HA is enabled or when cluster is 
upgraded and then namenode HA is enabled as in both the scenarios JournalNode 
and ZKFC service components are not created and wizard tries to install them.

We should make NameNode HA wizard more resilient. Let namenode HA wizard make 
API call to create service components for JournaNode and ZKFC as before. In 
case if we receive POST conflict error then web-client should not display it in 
an error pop-up and start installing components
{code}
{
  "status" : 409,
  "message" : "Attempted to create a component which already exists: 
[clusterName=c1, serviceName=HDFS, componentName=JOURNALNODE]"
}
{code}


  was:
In Ambari-1.6.0 release, we did not created JournalNode and ZKFC service 
components while installing a cluster and created it in NameNode HA wizard.

In Ambari-1.6.1 dev cycle we started creating all service components for all 
selected services including JournalNode and ZKFC. With BUG-18689 commit, we 
then removed creating JournalNoe and ZKFC in NameNode HA wizard. This creates 
problem when cluster is installed via blueprint and then namennode HA is 
enabled or when cluster is upgraded and then namenode HA is enabled as in both 
the scenarios JournalNode and ZKFC service components are not created and 
wizard tries to install them.

We should make NameNode HA wizard more resilient. Let namenode HA wizard make 
API call to create service components for JournaNode and ZKFC as before. In 
case if we receive POST conflict error then web-client should not display it in 
an error pop-up and start installing components
{code}
{
  "status" : 409,
  "message" : "Attempted to create a component which already exists: 
[clusterName=c1, serviceName=HDFS, componentName=JOURNALNODE]"
}
{code}



> NameNode HA wizard: create JournalNode and ZKFC components before installing 
> them
> ---------------------------------------------------------------------------------
>
>                 Key: AMBARI-6468
>                 URL: https://issues.apache.org/jira/browse/AMBARI-6468
>             Project: Ambari
>          Issue Type: Task
>          Components: client
>    Affects Versions: 1.7.0
>            Reporter: Antonenko Alexander
>            Assignee: Antonenko Alexander
>             Fix For: 1.7.0
>
>         Attachments: AMBARI-6468.patch
>
>
> In Ambari-1.6.0 release, we did not created JournalNode and ZKFC service 
> components while installing a cluster and created it in NameNode HA wizard.
> In Ambari-1.6.1 dev cycle we started creating all service components for all 
> selected services including JournalNode and ZKFC. After we removed creating 
> JournalNoe and ZKFC in NameNode HA wizard. This creates problem when cluster 
> is installed via blueprint and then namennode HA is enabled or when cluster 
> is upgraded and then namenode HA is enabled as in both the scenarios 
> JournalNode and ZKFC service components are not created and wizard tries to 
> install them.
> We should make NameNode HA wizard more resilient. Let namenode HA wizard make 
> API call to create service components for JournaNode and ZKFC as before. In 
> case if we receive POST conflict error then web-client should not display it 
> in an error pop-up and start installing components
> {code}
> {
>   "status" : 409,
>   "message" : "Attempted to create a component which already exists: 
> [clusterName=c1, serviceName=HDFS, componentName=JOURNALNODE]"
> }
> {code}



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Reply via email to