Hi David,
You can set the member expiry timeout in
Stratos-Home/repository/conf/autoscaler.xml via expiryTimeout element
(Highlighted below)
<configuration>
<autoscaler>
<rulesEvaluator>
<schedule>
<initialDelay>30</initialDelay>
<period>15</period>
</schedule>
</rulesEvaluator>
<cloudController>
<hostname>localhost</hostname>
<port>9444</port>
<!-->CC client timout in ms<-->
<clientTimeout>300000</clientTimeout>
</cloudController>
<stratosManager>
<hostname>localhost</hostname>
<port>9445</port>
<!--SM client timeout in ms-->
<clientTimeout>300000</clientTimeout>
</stratosManager>
<member>
<!-->member expiry timeout (time waited till activation) in
ms<-->
*<expiryTimeout>900000</expiryTimeout>*
</member>
<!-- monitor task interval -->
<monitorInterval>90000</monitorInterval>
</autoscaler>
</configuration>
Thanks.
On Mon, Mar 9, 2015 at 7:56 PM, david hbase <[email protected]> wrote:
> Hi, Imesh and Lakmal
>
> Thanks for you support, I will get more detail debug logs.
> An further issue is how can we disable / change the Member expriy time?
>
> [2015-03-08 09:41:10,723] DEBUG
> {org.apache.stratos.autoscaler.PartitionContext} - Member expiry time is
> set to: 900000
>
> Thanks
> -D
>
> On Mon, Mar 9, 2015 at 2:35 AM, Lakmal Warusawithana <[email protected]>
> wrote:
>
>>
>>
>> On Mon, Mar 9, 2015 at 2:39 PM, Imesh Gunaratne <[email protected]> wrote:
>>
>>> Hi Lakmal,
>>>
>>> On Mon, Mar 9, 2015 at 12:20 PM, Lakmal Warusawithana <[email protected]>
>>> wrote:
>>>>
>>>>
>>>> On Mon, Mar 9, 2015 at 12:16 PM, Imesh Gunaratne <[email protected]>
>>>> wrote:
>>>>
>>>>> Hi David,
>>>>>
>>>>> We might need to check why the load balancer is not sending statistics
>>>>> to CEP. Try to enable debug logs in the load balancer and run, it should
>>>>> trace the statistics sent to CEP.
>>>>>
>>>>> No AFAIK, it is not possible to configure Autoscaler to disregard load
>>>>> balancer statistics. I see this as a very valid requirement (to exclude
>>>>> any
>>>>> of the scaling factors), may be we can consider adding this to 4.1.0.
>>>>> Would
>>>>> you mind reporting this as a feature request?
>>>>>
>>>>>
>>>> I'm bit confuse here! I can scale up a cluster without having a LB. (in
>>>> 4.1.0) which mean LB is not a MUST right?
>>>>
>>>
>>> Yes, scaling up would work but scaling down might not work if statistics
>>> for all the autoscaling factors are available:
>>>
>>> scaleUp : Boolean() from ((rifReset && (rifPredictedValue >
>>> rifAverageLimit * 0.8)) || (mcReset && (mcPredictedValue > mcAverageLimit *
>>> 0.8)) || (laReset && (laPredictedValue > laAverageLimit * 0.8)))
>>> scaleDown : Boolean() from ((rifReset && (rifPredictedValue <
>>> rifAverageLimit * 0.1)) && (mcReset && (mcPredictedValue < mcAverageLimit *
>>> 0.1)) && (laReset && (laPredictedValue < laAverageLimit * 0.1)))
>>>
>>>
>>> https://github.com/apache/stratos/blob/4.0.0/products/stratos/modules/distribution/src/main/conf/scaling.drl
>>>
>>>
>> IMO, we should fix this. We should support scale up/down with selected
>> autoscaling parameters.
>>
>>
>>
>>> Thanks
>>>
>>>
>>
>>
>> --
>> Lakmal Warusawithana
>> Vice President, Apache Stratos
>> Director - Cloud Architecture; WSO2 Inc.
>> Mobile : +94714289692
>> Blog : http://lakmalsview.blogspot.com/
>>
>>
>
--
Rajkumar Rajaratnam
Committer & PMC Member, Apache Stratos
Software Engineer, WSO2
Mobile : +94777568639
Blog : rajkumarr.com
