[ 
https://issues.apache.org/jira/browse/TRAFODION-2384?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15718436#comment-15718436
 ] 

ASF GitHub Bot commented on TRAFODION-2384:
-------------------------------------------

Github user selvaganesang commented on a diff in the pull request:

    https://github.com/apache/incubator-trafodion/pull/862#discussion_r90761237
  
    --- Diff: install/python-installer/traf_setup.py ---
    @@ -50,6 +50,9 @@ def run():
         ### kernel settings ###
         run_cmd('sysctl -w kernel.pid_max=65535 2>&1 > /dev/null')
         run_cmd('echo "kernel.pid_max=65535" >> /etc/sysctl.conf')
    +    run_cmd('cp $SQ_ROOT/sysinstall/etc/init.d/trafodion /etc/init.d')
    --- End diff --
    
    Good point. I would leave it as is till /etc/trafodion/trafodion_config is 
available.  The service script needs to be executed by root as part of 
shutdown.  The variable SQ_ROOT will not be available to root user unless the 
service script trafodion is regenerated on the fly after installation or the 
new installer stores some default info at /etc/default/trafodion
    
    Also, by leaving it as it is possible for the existing installation to 
configure trafodion as a service by issuing the commands in the traf_setup.py 
manually


> Trafodion node needs to be gracefully shutdown when the node is brought down
> ----------------------------------------------------------------------------
>
>                 Key: TRAFODION-2384
>                 URL: https://issues.apache.org/jira/browse/TRAFODION-2384
>             Project: Apache Trafodion
>          Issue Type: Improvement
>          Components: installer
>            Reporter: Selvaganesan Govindarajan
>            Assignee: Selvaganesan Govindarajan
>             Fix For: 2.1-incubating
>
>
> When a node in a Trafodion cluster is brought down using linux commands like 
> shutdown, init 6 and reboot, the trafodion processes need to be stopped 
> before the underlying the stack is stopped. Otherwise, some of the processes 
> can dump core which are being restarted as part of the persistence concept of 
> Trafodion. RMS processes, TM are examples of such persistence processes.



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

Reply via email to