You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@stratos.apache.org by Rajkumar Rajaratnam <ra...@wso2.com> on 2014/10/14 11:42:25 UTC

Is there any problem with restarting/crashing Stratos?

Hi Devs,

Please be kind enough to explain me if I understood it wrong.

Topology is updated after every member is created. Stratos can go
down/crash after a member is created but before topology is updated. When
we start the stratos again, AS reads the topology and populate the member
lists. But topology doesn't have this member, so AS does not know about
this member.

With current logic CEP cannot identify this member as a faulty one, because
member is not in the topology, so CEP will reject any events from this
member.

So this member is going to stay there in the IaaS.

Is this a possible scenario?

We can make CEP to identify this as a faulty member as below;
If a member sends events to CEP and the member is not in the topology, then
CEP can complain it as a faulty member. This solution can work if the
member is sending events to CEP. There can be situation where cartridge
agent can crash. Then it will not send any events to CEP, so CEP also
doesn't know about this member.

How can we solve this?

Thanks.

-- 
Rajkumar Rajaratnam
Software Engineer | WSO2, Inc.
Mobile +94777568639 | +94783498120

Re: Is there any problem with restarting/crashing Stratos?

Posted by Manula Chathurika Thantriwatte <ma...@wso2.com>.
Hi,

This might be a rare scenario. I think according to your current design we
are unable to track such incidents.

On Tue, Oct 14, 2014 at 3:30 PM, Udara Liyanage <ud...@wso2.com> wrote:

> Hi,
>
> Even CEP mark the instance as faulty, CC won't be able to kill it since it
> does not have a reference to the instance.
>
> On Tue, Oct 14, 2014 at 3:12 PM, Rajkumar Rajaratnam <ra...@wso2.com>
> wrote:
>
>> Hi Devs,
>>
>> Please be kind enough to explain me if I understood it wrong.
>>
>> Topology is updated after every member is created. Stratos can go
>> down/crash after a member is created but before topology is updated. When
>> we start the stratos again, AS reads the topology and populate the member
>> lists. But topology doesn't have this member, so AS does not know about
>> this member.
>>
>> With current logic CEP cannot identify this member as a faulty one,
>> because member is not in the topology, so CEP will reject any events from
>> this member.
>>
>> So this member is going to stay there in the IaaS.
>>
>> Is this a possible scenario?
>>
>> We can make CEP to identify this as a faulty member as below;
>> If a member sends events to CEP and the member is not in the topology,
>> then CEP can complain it as a faulty member. This solution can work if the
>> member is sending events to CEP. There can be situation where cartridge
>> agent can crash. Then it will not send any events to CEP, so CEP also
>> doesn't know about this member.
>>
>> How can we solve this?
>>
>> Thanks.
>>
>> --
>> Rajkumar Rajaratnam
>> Software Engineer | WSO2, Inc.
>> Mobile +94777568639 | +94783498120
>>
>
>
>
> --
>
> Udara Liyanage
> Software Engineer
> WSO2, Inc.: http://wso2.com
> lean. enterprise. middleware
>
> web: http://udaraliyanage.wordpress.com
> phone: +94 71 443 6897
>



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

email : manulac@wso2.com / manula@apache.org
phone : +94 772492511
blog : http://manulachathurika.blogspot.com/

Re: Is there any problem with restarting/crashing Stratos?

Posted by Udara Liyanage <ud...@wso2.com>.
Hi,

Even CEP mark the instance as faulty, CC won't be able to kill it since it
does not have a reference to the instance.

On Tue, Oct 14, 2014 at 3:12 PM, Rajkumar Rajaratnam <ra...@wso2.com>
wrote:

> Hi Devs,
>
> Please be kind enough to explain me if I understood it wrong.
>
> Topology is updated after every member is created. Stratos can go
> down/crash after a member is created but before topology is updated. When
> we start the stratos again, AS reads the topology and populate the member
> lists. But topology doesn't have this member, so AS does not know about
> this member.
>
> With current logic CEP cannot identify this member as a faulty one,
> because member is not in the topology, so CEP will reject any events from
> this member.
>
> So this member is going to stay there in the IaaS.
>
> Is this a possible scenario?
>
> We can make CEP to identify this as a faulty member as below;
> If a member sends events to CEP and the member is not in the topology,
> then CEP can complain it as a faulty member. This solution can work if the
> member is sending events to CEP. There can be situation where cartridge
> agent can crash. Then it will not send any events to CEP, so CEP also
> doesn't know about this member.
>
> How can we solve this?
>
> Thanks.
>
> --
> Rajkumar Rajaratnam
> Software Engineer | WSO2, Inc.
> Mobile +94777568639 | +94783498120
>



-- 

Udara Liyanage
Software Engineer
WSO2, Inc.: http://wso2.com
lean. enterprise. middleware

web: http://udaraliyanage.wordpress.com
phone: +94 71 443 6897