You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@helix.apache.org by Bo Liu <ne...@gmail.com> on 2018/02/15 23:18:59 UTC

Throttle at partition level?

Hi,

Does Helix support throttling state transition at partition level?
I only find cluster, resource and instance level throttling as below:

public enum ThrottleScope {
  CLUSTER,
  RESOURCE,
  INSTANCE
}



-- 
Best regards,
Bo

Re: Throttle at partition level?

Posted by Bo Liu <ne...@gmail.com>.

Hi Lei,

Yes, we'd like at most one transition for a partition.

Besides this, we also need instance and cluster level throttling. Can you
take a look at the configuration I posted above. It didn't work for us.
Anything we missed?

Thanks,
Bo

On Feb 16, 2018 08:50, "Lei Xia" <xi...@gmail.com> wrote:

> Hi, Bo
>
>   Can you elaborate more on your use case?  Are you saying you do not want
> any offline->slave transition if there is an ongoing Master->Slave
> transition?
>
>
>
> Lei
>
> On Fri, Feb 16, 2018 at 8:20 AM, kishore g <g....@gmail.com> wrote:
>
>> I thought we had throttling support at the partition scope. I will have
>> to check the code and get back to you
>>
>> On Fri, Feb 16, 2018 at 7:55 AM, Bo Liu <ne...@gmail.com> wrote:
>>
>>> Your input on how to configure Helix for Throttling is appreciated.
>>>
>>> On Thu, Feb 15, 2018 at 5:09 PM, Bo Liu <ne...@gmail.com> wrote:
>>>
>>>> A related question. Is this the proper way to enable instance and
>>>> cluster level throttling?
>>>> Basically we want to allow up to 100 concurrent state transitions at
>>>> cluster level and 2 at instance level.
>>>> This is configured as list fields of cluster configuration.
>>>> It didn't work. So I am wondering what's the proper way to configure it?
>>>>
>>>> Thanks,
>>>> Bo
>>>>
>>>> On Thu, Feb 15, 2018 at 3:18 PM, Bo Liu <ne...@gmail.com> wrote:
>>>>
>>>>> Hi,
>>>>>
>>>>> Does Helix support throttling state transition at partition level?
>>>>> I only find cluster, resource and instance level throttling as below:
>>>>>
>>>>> public enum ThrottleScope {
>>>>>   CLUSTER,
>>>>>   RESOURCE,
>>>>>   INSTANCE
>>>>> }
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Best regards,
>>>>> Bo
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Best regards,
>>>> Bo
>>>>
>>>>
>>>
>>>
>>> --
>>> Best regards,
>>> Bo
>>>
>>>
>>
>
>
> --
> Lei Xia
>

Re: Throttle at partition level?

Posted by Lei Xia <xi...@gmail.com>.

Hi, Bo

  Can you elaborate more on your use case?  Are you saying you do not want
any offline->slave transition if there is an ongoing Master->Slave
transition?



Lei

On Fri, Feb 16, 2018 at 8:20 AM, kishore g <g....@gmail.com> wrote:

> I thought we had throttling support at the partition scope. I will have
> to check the code and get back to you
>
> On Fri, Feb 16, 2018 at 7:55 AM, Bo Liu <ne...@gmail.com> wrote:
>
>> Your input on how to configure Helix for Throttling is appreciated.
>>
>> On Thu, Feb 15, 2018 at 5:09 PM, Bo Liu <ne...@gmail.com> wrote:
>>
>>> A related question. Is this the proper way to enable instance and
>>> cluster level throttling?
>>> Basically we want to allow up to 100 concurrent state transitions at
>>> cluster level and 2 at instance level.
>>> This is configured as list fields of cluster configuration.
>>> It didn't work. So I am wondering what's the proper way to configure it?
>>>
>>> Thanks,
>>> Bo
>>>
>>> On Thu, Feb 15, 2018 at 3:18 PM, Bo Liu <ne...@gmail.com> wrote:
>>>
>>>> Hi,
>>>>
>>>> Does Helix support throttling state transition at partition level?
>>>> I only find cluster, resource and instance level throttling as below:
>>>>
>>>> public enum ThrottleScope {
>>>>   CLUSTER,
>>>>   RESOURCE,
>>>>   INSTANCE
>>>> }
>>>>
>>>>
>>>>
>>>> --
>>>> Best regards,
>>>> Bo
>>>>
>>>>
>>>
>>>
>>> --
>>> Best regards,
>>> Bo
>>>
>>>
>>
>>
>> --
>> Best regards,
>> Bo
>>
>>
>


-- 
Lei Xia

Re: Throttle at partition level?

Posted by kishore g <g....@gmail.com>.

I thought we had throttling support at the partition scope. I will have to
check the code and get back to you

On Fri, Feb 16, 2018 at 7:55 AM, Bo Liu <ne...@gmail.com> wrote:

> Your input on how to configure Helix for Throttling is appreciated.
>
> On Thu, Feb 15, 2018 at 5:09 PM, Bo Liu <ne...@gmail.com> wrote:
>
>> A related question. Is this the proper way to enable instance and cluster
>> level throttling?
>> Basically we want to allow up to 100 concurrent state transitions at
>> cluster level and 2 at instance level.
>> This is configured as list fields of cluster configuration.
>> It didn't work. So I am wondering what's the proper way to configure it?
>>
>> Thanks,
>> Bo
>>
>> On Thu, Feb 15, 2018 at 3:18 PM, Bo Liu <ne...@gmail.com> wrote:
>>
>>> Hi,
>>>
>>> Does Helix support throttling state transition at partition level?
>>> I only find cluster, resource and instance level throttling as below:
>>>
>>> public enum ThrottleScope {
>>>   CLUSTER,
>>>   RESOURCE,
>>>   INSTANCE
>>> }
>>>
>>>
>>>
>>> --
>>> Best regards,
>>> Bo
>>>
>>>
>>
>>
>> --
>> Best regards,
>> Bo
>>
>>
>
>
> --
> Best regards,
> Bo
>
>

Re: Throttle at partition level?

Posted by Bo Liu <ne...@gmail.com>.

Your input on how to configure Helix for Throttling is appreciated.

On Thu, Feb 15, 2018 at 5:09 PM, Bo Liu <ne...@gmail.com> wrote:

> A related question. Is this the proper way to enable instance and cluster
> level throttling?
> Basically we want to allow up to 100 concurrent state transitions at
> cluster level and 2 at instance level.
> This is configured as list fields of cluster configuration.
> It didn't work. So I am wondering what's the proper way to configure it?
>
> Thanks,
> Bo
>
> On Thu, Feb 15, 2018 at 3:18 PM, Bo Liu <ne...@gmail.com> wrote:
>
>> Hi,
>>
>> Does Helix support throttling state transition at partition level?
>> I only find cluster, resource and instance level throttling as below:
>>
>> public enum ThrottleScope {
>>   CLUSTER,
>>   RESOURCE,
>>   INSTANCE
>> }
>>
>>
>>
>> --
>> Best regards,
>> Bo
>>
>>
>
>
> --
> Best regards,
> Bo
>
>


-- 
Best regards,
Bo

Re: Throttle at partition level?

Posted by Bo Liu <ne...@gmail.com>.

A related question. Is this the proper way to enable instance and cluster
level throttling?
Basically we want to allow up to 100 concurrent state transitions at
cluster level and 2 at instance level.
This is configured as list fields of cluster configuration.
It didn't work. So I am wondering what's the proper way to configure it?

Thanks,
Bo

On Thu, Feb 15, 2018 at 3:18 PM, Bo Liu <ne...@gmail.com> wrote:

> Hi,
>
> Does Helix support throttling state transition at partition level?
> I only find cluster, resource and instance level throttling as below:
>
> public enum ThrottleScope {
>   CLUSTER,
>   RESOURCE,
>   INSTANCE
> }
>
>
>
> --
> Best regards,
> Bo
>
>

-- 
Best regards,
Bo

Re: Throttle at partition level?

Posted by Bo Liu <ne...@gmail.com>.

It can help us simplify Master-Slave state transition.

When we bring up a new Slave, we need to set it to replicate from the
Master. If there is a concurrent Master->Slave transition, the thing starts
to get more complicated.

On Thu, Feb 15, 2018 at 5:12 PM, kishore g <g....@gmail.com> wrote:

> Can you describe the use case where you need throttling at a partition
> level.
>
> On Thu, Feb 15, 2018 at 3:18 PM, Bo Liu <ne...@gmail.com> wrote:
>
>> Hi,
>>
>> Does Helix support throttling state transition at partition level?
>> I only find cluster, resource and instance level throttling as below:
>>
>> public enum ThrottleScope {
>>   CLUSTER,
>>   RESOURCE,
>>   INSTANCE
>> }
>>
>>
>>
>> --
>> Best regards,
>> Bo
>>
>>
>

-- 
Best regards,
Bo

Re: Throttle at partition level?

Posted by kishore g <g....@gmail.com>.

Can you describe the use case where you need throttling at a partition
level.

On Thu, Feb 15, 2018 at 3:18 PM, Bo Liu <ne...@gmail.com> wrote:

> Hi,
>
> Does Helix support throttling state transition at partition level?
> I only find cluster, resource and instance level throttling as below:
>
> public enum ThrottleScope {
>   CLUSTER,
>   RESOURCE,
>   INSTANCE
> }
>
>
>
> --
> Best regards,
> Bo
>
>