AWS Load Balancer Support - issue with topology update

[Swapnil Patil <sw...@gmail.com>]

Hi All,

Till now I am able to create load balancer for a new cluster and attach its
members to load balancer.

But I am facing an issue.

I am using sample ec2 scripts to deploy application.

For initial topology update all works fine. I get a new cluster and member
in it. So the extension create load balancer and attaches member to it.

But in some time that member gets shut down and a new member gets started.
Now in the next topology update i still get the old instance id as a member.

This goes on and this second instance gets terminated and a new one is
started. But in next topology update, extension gets second instance id as
a member which is no longer a member.

Just wondering how to tackle this problem so that extension will get
correct topology update.

Thanks and Regards,
Swapnil

Re: AWS Load Balancer Support - issue with topology update

[Swapnil Patil <sw...@gmail.com>]

Thanks Akila!

Regards,
Swapnil

On Thu, Jul 16, 2015 at 2:11 PM, Akila Ravihansa Perera <ra...@wso2.com>
wrote:

> Hi Swapnil,
>
> There is a known issue in Topology model where if a member is detected as
> faulty by CEP, it will not remove that member from the topology until CC
> has terminated the instance (from IaaS level). If the instance is already
> terminated (manually or by other means), it will keep trying for a bounded
> time period and remove it from the topology.
>
> Until CC terminates or forcefully remove the instance, you will still see
> the member as active. This is a known problem and will be fixed in a future
> release. For now, you can listen to FaultyMemberEvents and remove those
> instances from the active member list that you maintain.
>
> On the side note, you should not be listening to topology updates except
> for the very first time. Initialize the member list from the first topology
> update and build the member list by listening to other member events on the
> fly.
>
> Thanks.
>
> On Thu, Jul 16, 2015 at 1:40 PM, Swapnil Patil <sw...@gmail.com>
> wrote:
>
>> Hi All,
>>
>> Till now I am able to create load balancer for a new cluster and attach
>> its members to load balancer.
>>
>> But I am facing an issue.
>>
>> I am using sample ec2 scripts to deploy application.
>>
>> For initial topology update all works fine. I get a new cluster and
>> member in it. So the extension create load balancer and attaches member to
>> it.
>>
>> But in some time that member gets shut down and a new member gets
>> started. Now in the next topology update i still get the old instance id as
>> a member.
>>
>> This goes on and this second instance gets terminated and a new one is
>> started. But in next topology update, extension gets second instance id as
>> a member which is no longer a member.
>>
>> Just wondering how to tackle this problem so that extension will get
>> correct topology update.
>>
>> Thanks and Regards,
>> Swapnil
>>
>
>
>
> --
> Akila Ravihansa Perera
> Software Engineer, WSO2
>
> Blog: http://ravihansa3000.blogspot.com
>

Re: AWS Load Balancer Support - issue with topology update

[Akila Ravihansa Perera <ra...@wso2.com>]

Hi Swapnil,

There is a known issue in Topology model where if a member is detected as
faulty by CEP, it will not remove that member from the topology until CC
has terminated the instance (from IaaS level). If the instance is already
terminated (manually or by other means), it will keep trying for a bounded
time period and remove it from the topology.

Until CC terminates or forcefully remove the instance, you will still see
the member as active. This is a known problem and will be fixed in a future
release. For now, you can listen to FaultyMemberEvents and remove those
instances from the active member list that you maintain.

On the side note, you should not be listening to topology updates except
for the very first time. Initialize the member list from the first topology
update and build the member list by listening to other member events on the
fly.

Thanks.

On Thu, Jul 16, 2015 at 1:40 PM, Swapnil Patil <sw...@gmail.com>
wrote:

> Hi All,
>
> Till now I am able to create load balancer for a new cluster and attach
> its members to load balancer.
>
> But I am facing an issue.
>
> I am using sample ec2 scripts to deploy application.
>
> For initial topology update all works fine. I get a new cluster and member
> in it. So the extension create load balancer and attaches member to it.
>
> But in some time that member gets shut down and a new member gets started.
> Now in the next topology update i still get the old instance id as a member.
>
> This goes on and this second instance gets terminated and a new one is
> started. But in next topology update, extension gets second instance id as
> a member which is no longer a member.
>
> Just wondering how to tackle this problem so that extension will get
> correct topology update.
>
> Thanks and Regards,
> Swapnil
>



-- 
Akila Ravihansa Perera
Software Engineer, WSO2

Blog: http://ravihansa3000.blogspot.com