[ 
https://issues.apache.org/jira/browse/HBASE-15523?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15496793#comment-15496793
 ] 

Dima Spivak commented on HBASE-15523:
-------------------------------------

Moving this to a variable (that would ostensibly need to be set on every 
machine in a cluster) moves the onus from the script to the user and also means 
you couldn't grep your cluster's restart behavior via looking at the process 
tree. For those reasons, I'm not a fan of this proposal (and I'm still not 
clear on why it was reopened after being closed for six months).

> enhance hbase-daemon.sh to enable autorestart.
> ----------------------------------------------
>
>                 Key: HBASE-15523
>                 URL: https://issues.apache.org/jira/browse/HBASE-15523
>             Project: HBase
>          Issue Type: Improvement
>            Reporter: Yi Liang
>            Assignee: Yi Liang
>            Priority: Minor
>         Attachments: HBASE-15523.patch
>
>
> enhance hbase-daemon.sh to enable autorestart.
> component(like master, region server) will auto-start when terminated/killed 
> abnormally if 
>            (a) Add a new env variable $HBASE_AUTORESTART to hbase-env.sh i.e. 
>          export HBASE_AUTORESTART=true
>           (b) Then add the following 3 simple lines(59-61) to  
> /bin/hbase-daemon.sh
>      
>      51 # get arguments
>      52 startStop=$1
>      53 shift
>      54
>      55 command=$1
>      56 shift
>      57
>      58 #make sure the auto-restart are default settings
>      59 if [ "$HBASE_AUTORESTART" == "true" ] && [ "$startStop" == "start" ]; 
> then     
>      60   startStop="autorestart"     
>      61 fi



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

Reply via email to