You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@storm.apache.org by yang hui <lo...@163.com> on 2014/08/05 04:55:22 UTC
how can I make the Nimbus highly available ?
HI,all:
how can I make the Nimbus highly available ? I am just confused about this. In the website, I just read below;
"This means you can kill -9 Nimbus or the Supervisors and they’ll start back up like nothing happened. This design leads to Storm clusters being incredibly stable."
But if my master node crashes with an unexpected hardware failure , I can't recover it in a short time. So how can I make the Nimbus as a high available service ?
Re: how can I make the Nimbus highly available ?
Posted by yang hui <lo...@163.com>.
Thanks a lot.
On 2014-8-6, at 上午2:18, "P. Taylor Goetz" <pt...@gmail.com> wrote:
> HA for Nimbus is planned, and there has already been some work in that area:
>
> https://issues.apache.org/jira/browse/STORM-166
>
> Nimbus is not a true SPOF in that if it goes down, you’re topologies will continue to function and process data. With Nimbus down what you lose is the ability to submit/manage topologies and reassign tasks.
>
> Without STORM-166, you would have to manually intervene if your nimbus host exploded. If it is a big concern you may want a standby machine and something in your ops pocket to quickly change DNS for your nimbus host.
>
> - Taylor
>
> On Aug 4, 2014, at 10:55 PM, yang hui <lo...@163.com> wrote:
>
>> HI,all:
>> how can I make the Nimbus highly available ? I am just confused about this. In the website, I just read below;
>> "This means you can kill -9 Nimbus or the Supervisors and they’ll start back up like nothing happened. This design leads to Storm clusters being incredibly stable."
>>
>> But if my master node crashes with an unexpected hardware failure , I can't recover it in a short time. So how can I make the Nimbus as a high available service ?
>>
>
> 1 attachments
> signature.asc(1K)
> download preview
Re: how can I make the Nimbus highly available ?
Posted by "P. Taylor Goetz" <pt...@gmail.com>.
HA for Nimbus is planned, and there has already been some work in that area:
https://issues.apache.org/jira/browse/STORM-166
Nimbus is not a true SPOF in that if it goes down, you’re topologies will continue to function and process data. With Nimbus down what you lose is the ability to submit/manage topologies and reassign tasks.
Without STORM-166, you would have to manually intervene if your nimbus host exploded. If it is a big concern you may want a standby machine and something in your ops pocket to quickly change DNS for your nimbus host.
- Taylor
On Aug 4, 2014, at 10:55 PM, yang hui <lo...@163.com> wrote:
> HI,all:
> how can I make the Nimbus highly available ? I am just confused about this. In the website, I just read below;
> "This means you can kill -9 Nimbus or the Supervisors and they’ll start back up like nothing happened. This design leads to Storm clusters being incredibly stable."
>
> But if my master node crashes with an unexpected hardware failure , I can't recover it in a short time. So how can I make the Nimbus as a high available service ?
>