How to remove or replace a worker

[Dillian Murphey <cr...@gmail.com>]

Does it make sense to have a procedure to replace a worker node without
shutting down the entire cluster?

I want to replace a worker node with no impact to the currently running
processes. Does that make sense in a storm context (noob here).

I've learned that stopping the supervisor process is not enough. The
workers will continue to run.  Can I kill those workers? Will the work get
automatically redistributed?

I basically, want to replace one worker node server with another with no
impact.

Thanks for any help. Been looking around for the past 20 mins and not
finding a good match to my question.

Re: How to remove or replace a worker

[Dillian Murphey <cr...@gmail.com>]

Thanks for confirming, John.

On Tue, Jun 30, 2015 at 10:41 AM, John Reilly <jr...@inconspicuous.org> wrote:

> If you shutdown the supervisor on a host and then kill the worker
> processes on the host, then they should get restarted on other worker hosts.
>
> On Mon, Jun 29, 2015 at 5:04 PM Dillian Murphey <cr...@gmail.com>
> wrote:
>
>> Does it make sense to have a procedure to replace a worker node without
>> shutting down the entire cluster?
>>
>> I want to replace a worker node with no impact to the currently running
>> processes. Does that make sense in a storm context (noob here).
>>
>> I've learned that stopping the supervisor process is not enough. The
>> workers will continue to run.  Can I kill those workers? Will the work get
>> automatically redistributed?
>>
>> I basically, want to replace one worker node server with another with no
>> impact.
>>
>> Thanks for any help. Been looking around for the past 20 mins and not
>> finding a good match to my question.
>>
>>

Re: How to remove or replace a worker

[John Reilly <jr...@inconspicuous.org>]

If you shutdown the supervisor on a host and then kill the worker processes
on the host, then they should get restarted on other worker hosts.

On Mon, Jun 29, 2015 at 5:04 PM Dillian Murphey <cr...@gmail.com>
wrote:

> Does it make sense to have a procedure to replace a worker node without
> shutting down the entire cluster?
>
> I want to replace a worker node with no impact to the currently running
> processes. Does that make sense in a storm context (noob here).
>
> I've learned that stopping the supervisor process is not enough. The
> workers will continue to run.  Can I kill those workers? Will the work get
> automatically redistributed?
>
> I basically, want to replace one worker node server with another with no
> impact.
>
> Thanks for any help. Been looking around for the past 20 mins and not
> finding a good match to my question.
>
>