Hi Subash,

On Fri, Feb 21, 2014 at 5:43 PM, Subash Chaturanga <[email protected]> wrote:

> HI Janaka,
> In  a Prod or non Prod env ithere can be a use case that at a given point
> of time, governance master needs to refactor the LC config (some config
> change as Ajith mentioned) and reflect the changes on all associated 100s
> of services. So I believe this feature is a useful one (may be in a new
> form like "refactor").
>
But the current behavior of G-Reg is not to update the existing artifacts.
Is that a design decision?

Thanks,
Janaka

>
> Vendors like WSRR have the capability to maintain versions of the same LC,
> which is a good feature to have for GReg as well, where it will allows
> users to create versions of LC for those major/minor changes and it will
> avoid  those inconsistencies.
>
>
>
>
> On Sat, Feb 22, 2014 at 3:51 AM, Janaka Ranabahu <[email protected]> wrote:
>
>> Hi Ajith,
>>
>> Please see my concerns inline.
>>
>>
>> On Fri, Feb 21, 2014 at 4:33 PM, Ajith Vitharana <[email protected]> wrote:
>>
>>> Hi Janaka,
>>>
>>> First *"**ServiceLifeCycle is already in use and your changes may
>>> affect existing usage" * is sufficient warning. Because we can't add
>>> lengthy descriptions in small UI box.
>>> Practically, do we change the *LC state id* having large number of
>>> artifacts in a production system ?.
>>>
>> What happens if some user do so? This may not happen in a production
>> instance. But in a developer environment this is something very likely to
>> happen.
>>
>>>
>>> Renaming the  existing *LC state id* is only one use case.
>>>
>> yes it is only one possibility. It is the first one I tried and it
>> created couple of inconsistent behaviors. Also we can not limit users to
>> not to change the state id. It is one possibility that we have allowed.
>>
>> Following are are the valid points of allowing to update the existing LC
>>> configuration.
>>>
>>> 1. Add new state to existing configurations.
>>>
>> 2. Add/update executors, validators, transition UI ..etc.
>>>
>> These 2 are not likely scenarios for a already defined LC. If there is a
>> requirement for a new state, that indicates that the organization process
>> has changed. Then what we need is a new version of the LC and not to modify
>> the current one.
>>
>>> 3. Add/Update the parameters.
>>> 4 Add/update the check Items ..etc
>>>
>> This will not work for the existing resources. I tried this and it was
>> not reflected immediately. So that is why I raised the concern that by
>> allowing this, there can be resources in the same state but with different
>> check items.
>>
>> So IMO we should not allow to modify a LC if it is already in use. By not
>> allowing, we can enforce the users to have a naming conversion for the LC,
>> which is version aware and there is very limited opportunity for
>> inconsistent behavior.
>>
>> WDYT?
>>
>> Thanks,
>> Janaka
>>
>>>
>>> Thanks.
>>> Ajith.
>>>
>>>
>>>
>>> On Sat, Feb 22, 2014 at 2:16 AM, Janaka Ranabahu <[email protected]>wrote:
>>>
>>>> Hi,
>>>>
>>>> I have noticed that we allow users to modify a lifecycle that is
>>>> already assigned to resources. Did we do a complete analysis of the
>>>> situations that could come by allowing this behavior?
>>>>
>>>> I have noticed the following inconsistent behavior due to this.
>>>>
>>>> When saving the lifecycle configuration, it only gives a warning saying*
>>>> "ServiceLifeCycle is already in use and your changes may affect existing
>>>> usage. Are you sure you want to save these Lifecycle changes?"* There
>>>> is no descriptive explanation about what is the impact of the change. It
>>>> happens to be that, if you rename a state, and if there were resources on
>>>> that state, the lifecycle portlet disappear from the UI. There is no way to
>>>> remove, add or do any action on the lifecycle for that resource and that
>>>> resource is completely unusable(from LC point of view) from that point
>>>> onward.
>>>>
>>>> Also we do not update the existing resources with the change. If we add
>>>> a new check item or change a check item of an existing state, that change
>>>> is not reflected for the resources that are already associated with the
>>>> lifecycle. Because of that, G-Reg began to have resources with the same
>>>> lifecycle, but with different behaviors.
>>>>
>>>> Therefore, I'm wondering what is the purpose of modifying a lifecycle
>>>> which is already in use. IMO, this had introduced a inconsistent behavior
>>>> and had not solve anything. The real problem comes when there is a high
>>>> number of resources that is associated with the lifecycle.
>>>>
>>>> Thanks,
>>>> Janaka
>>>>
>>>> --
>>>> *Janaka Ranabahu*
>>>> Senior Software Engineer; WSO2 Inc.; http://wso2.com
>>>>
>>>>
>>>> * E-mail: [email protected] <http://wso2.com>**M: **+94 718370861
>>>> <%2B94%20718370861>*
>>>>
>>>> Lean . Enterprise . Middleware
>>>>
>>>
>>>
>>>
>>> --
>>> Ajith Vitharana.
>>> WSO2 Inc. - http://wso2.org
>>> Email  :  [email protected]
>>> Mobile : +94772217350
>>>
>>>
>>
>>
>> --
>> *Janaka Ranabahu*
>> Senior Software Engineer; WSO2 Inc.; http://wso2.com
>>
>>
>> * E-mail: [email protected] <http://wso2.com>**M: **+94 718370861
>> <%2B94%20718370861>*
>>
>> Lean . Enterprise . Middleware
>>
>
>
>
> --
> Thanks
> /subash
>
> *Subash Chaturanga*
> Senior Software Engineer :Integration TG; WSO2 Inc. http://wso2.com
>
> email: [email protected]
> blog:  http://subashsdm.blogspot.com/
> twitter: @subash89
> phone: +9477 2225922
> Lean . Enterprise . Middleware
>



-- 
*Janaka Ranabahu*
Senior Software Engineer; WSO2 Inc.; http://wso2.com


* E-mail: [email protected] <http://wso2.com>**M: **+94 718370861
<%2B94%20718370861>*

Lean . Enterprise . Middleware
_______________________________________________
Architecture mailing list
[email protected]
https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture

Reply via email to