Hi,

+1 for that changes. Thanks Raj for finding it.

Thanks !

On Tue, Nov 4, 2014 at 6:02 PM, Isuru Haththotuwa <[email protected]> wrote:

>
>
> On Tue, Nov 4, 2014 at 5:35 PM, Rajkumar Rajaratnam <[email protected]>
> wrote:
>
>> Hi,
>>
>> Noticed that we are calling CC API (once) to terminate an instance
>> directly on receiving some events. And we are assuming instances will be
>> terminated without any problem.
>>
>> It is a risk!
>>
>>    - We will get these events only once
>>    - We are calling CC API to terminate the instances
>>    - Say instance termination is failing for some reasons
>>
>> What is the situation now? AS will not ask CC to terminate these
>> instances again. Because, you will not get these events (say
>> ClusterTerminatingEvent) again. So these instances will never be terminated.
>>
>> IMO, on receiving these events, we should move these members to a member
>> list(may be obsolete member list) and let drools to terminate this
>> instances. When AS receives member terminated event for the member, we will
>> remove it from the list. This way, AS will ask CC to terminate the instance
>> until it gets member terminated event. This flow is already there for
>> obsoleted members. So only thing we have to do is to move these members to
>> obsolete member list when receiving these events.
>>
>> wdyt?
>>
> +1, great work on figuring out this potential issue Raj.
>
>> Thanks.
>>
>> --
>> Raj
>>
>> --
>> Thanks and Regards,
>>
>> Isuru H.
>> +94 716 358 048* <http://wso2.com/>*
>>
>>
>> * <http://wso2.com/>*
>>
>>
>>


-- 
Regards,
Manula Chathurika Thantriwatte
Software Engineer
WSO2 Inc. : http://wso2.com
lean . enterprise . middleware

email : [email protected] / [email protected]
phone : +94 772492511
blog : http://manulachathurika.blogspot.com/

Reply via email to