You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@stratos.apache.org by Lahiru Sandaruwan <la...@wso2.com> on 2013/12/10 04:46:43 UTC
Keeping the instance state in Topology
Hi,
Need to know about $subject.
These details are valuable for handling a restart of the Autoscaler.
I see Activated, Created, Starting, Suspended, and Terminated states.
I think we can map Created state to Pending state at Autoscaler side.
What is the status if the instance is created but not activated? Is it
starting?
Thanks.
--
--
Lahiru Sandaruwan
Software Engineer,
Platform Technologies,
WSO2 Inc., http://wso2.com
lean.enterprise.middleware
email: lahirus@wso2.com cell: (+94) 773 325 954
blog: http://lahiruwrites.blogspot.com/
twitter: http://twitter.com/lahirus
linked-in: http://lk.linkedin.com/pub/lahiru-sandaruwan/16/153/146
Re: Keeping the instance state in Topology
Posted by Udara Liyanage <ud...@wso2.com>.
Currently when AS started an instance, it keep that member in a pending
list.
private List<MemberContext> pendingMembers; in PartitionContext.
If the member did not become activated for some time, AS takes necessary
actions (kill it and start another I guess).
On Tue, Dec 10, 2013 at 2:01 AM, Reka Thirunavukkarasu <re...@wso2.com>wrote:
> Hi
>
>
> On Tue, Dec 10, 2013 at 9:16 AM, Lahiru Sandaruwan <la...@wso2.com>wrote:
>
>> Hi,
>>
>> Need to know about $subject.
>> These details are valuable for handling a restart of the Autoscaler.
>>
>> I see Activated, Created, Starting, Suspended, and Terminated states.
>>
>
>> I think we can map Created state to Pending state at Autoscaler side.
>>
> +1.
>
>>
>> What is the status if the instance is created but not activated? Is it
>> starting?
>>
> Yah. Started means our cartridge agent is ready. But if you don't receive
> the activated event after started event, that means the application inside
> the cartridge failed to start. We have to track this also in the
> autoscaler. Then only if the application is not active, we can kill the
> existing faulty instance and bring up a new one (in the mean time, we can
> assume that the faulty application gets fixed in the repository or in the
> cartridge image).
>
> This is just my thought. But Autoscaler can't keep doing this, if noone is
> there to fix the application. In that case, after few attempts, autoscaler
> has to give up until it notifies again, when the application gets fixed
> either via the repository or via the image or in any other way. WDYT?
>
> Thanks,
> Reka
>
>>
>> Thanks.
>>
>> --
>> --
>> Lahiru Sandaruwan
>> Software Engineer,
>> Platform Technologies,
>> WSO2 Inc., http://wso2.com
>> lean.enterprise.middleware
>>
>> email: lahirus@wso2.com cell: (+94) 773 325 954
>> blog: http://lahiruwrites.blogspot.com/
>> twitter: http://twitter.com/lahirus
>> linked-in: http://lk.linkedin.com/pub/lahiru-sandaruwan/16/153/146
>>
>>
>
>
> --
> Reka Thirunavukkarasu
> Software Engineer,
> WSO2, Inc.:http://wso2.com,
> Mobile: +94776442007
>
>
>
--
Udara Liyanage
Software Engineer
WSO2, Inc.: http://wso2.com
lean. enterprise. middleware
web: http://udaraliyanage.wordpress.com
phone: +94 71 443 6897
Re: Keeping the instance state in Topology
Posted by Lahiru Sandaruwan <la...@wso2.com>.
Good suggestions Reka. We might have to do this carefully.
Instance status kept in topology will be helpful to a recovery of
Autoscaler.
Thanks.
On Tue, Dec 10, 2013 at 12:31 PM, Reka Thirunavukkarasu <re...@wso2.com>wrote:
> Hi
>
>
> On Tue, Dec 10, 2013 at 9:16 AM, Lahiru Sandaruwan <la...@wso2.com>wrote:
>
>> Hi,
>>
>> Need to know about $subject.
>> These details are valuable for handling a restart of the Autoscaler.
>>
>> I see Activated, Created, Starting, Suspended, and Terminated states.
>>
>
>> I think we can map Created state to Pending state at Autoscaler side.
>>
> +1.
>
>>
>> What is the status if the instance is created but not activated? Is it
>> starting?
>>
> Yah. Started means our cartridge agent is ready. But if you don't receive
> the activated event after started event, that means the application inside
> the cartridge failed to start. We have to track this also in the
> autoscaler. Then only if the application is not active, we can kill the
> existing faulty instance and bring up a new one (in the mean time, we can
> assume that the faulty application gets fixed in the repository or in the
> cartridge image).
>
> This is just my thought. But Autoscaler can't keep doing this, if noone is
> there to fix the application. In that case, after few attempts, autoscaler
> has to give up until it notifies again, when the application gets fixed
> either via the repository or via the image or in any other way. WDYT?
>
> Thanks,
> Reka
>
>>
>> Thanks.
>>
>> --
>> --
>> Lahiru Sandaruwan
>> Software Engineer,
>> Platform Technologies,
>> WSO2 Inc., http://wso2.com
>> lean.enterprise.middleware
>>
>> email: lahirus@wso2.com cell: (+94) 773 325 954
>> blog: http://lahiruwrites.blogspot.com/
>> twitter: http://twitter.com/lahirus
>> linked-in: http://lk.linkedin.com/pub/lahiru-sandaruwan/16/153/146
>>
>>
>
>
> --
> Reka Thirunavukkarasu
> Software Engineer,
> WSO2, Inc.:http://wso2.com,
> Mobile: +94776442007
>
>
>
--
--
Lahiru Sandaruwan
Software Engineer,
Platform Technologies,
WSO2 Inc., http://wso2.com
lean.enterprise.middleware
email: lahirus@wso2.com cell: (+94) 773 325 954
blog: http://lahiruwrites.blogspot.com/
twitter: http://twitter.com/lahirus
linked-in: http://lk.linkedin.com/pub/lahiru-sandaruwan/16/153/146
Re: Keeping the instance state in Topology
Posted by Reka Thirunavukkarasu <re...@wso2.com>.
Hi
On Tue, Dec 10, 2013 at 9:16 AM, Lahiru Sandaruwan <la...@wso2.com> wrote:
> Hi,
>
> Need to know about $subject.
> These details are valuable for handling a restart of the Autoscaler.
>
> I see Activated, Created, Starting, Suspended, and Terminated states.
>
> I think we can map Created state to Pending state at Autoscaler side.
>
+1.
>
> What is the status if the instance is created but not activated? Is it
> starting?
>
Yah. Started means our cartridge agent is ready. But if you don't receive
the activated event after started event, that means the application inside
the cartridge failed to start. We have to track this also in the
autoscaler. Then only if the application is not active, we can kill the
existing faulty instance and bring up a new one (in the mean time, we can
assume that the faulty application gets fixed in the repository or in the
cartridge image).
This is just my thought. But Autoscaler can't keep doing this, if noone is
there to fix the application. In that case, after few attempts, autoscaler
has to give up until it notifies again, when the application gets fixed
either via the repository or via the image or in any other way. WDYT?
Thanks,
Reka
>
> Thanks.
>
> --
> --
> Lahiru Sandaruwan
> Software Engineer,
> Platform Technologies,
> WSO2 Inc., http://wso2.com
> lean.enterprise.middleware
>
> email: lahirus@wso2.com cell: (+94) 773 325 954
> blog: http://lahiruwrites.blogspot.com/
> twitter: http://twitter.com/lahirus
> linked-in: http://lk.linkedin.com/pub/lahiru-sandaruwan/16/153/146
>
>
--
Reka Thirunavukkarasu
Software Engineer,
WSO2, Inc.:http://wso2.com,
Mobile: +94776442007