Am 10.08.2015 um 19:44 schrieb Daniel Henrique Barboza:
On 08/10/2015 02:31 PM, Chandra Shehkhar Reddy Potula
chandra(a)linux.vnet.ibm.com [ginger-dev] wrote:
> Hi,
>
> It is good idea to modularize the host functionality as plugins.
> Keeping common host functionality as a one plugin and at the same
> time having platform specific host functionality as additional
> plugins gives choice of picking up the relevant features on specific
> platform.
>
> Since we had an agreement on the movement of host functionality from
> kimchi to ginger, here is draft proposal I have to modularize the
> ginger functionality.
>
> *Plugin ginger-basic:*
> Base Information
> Basic Info
> Host Stats
> Debug Reports
>
> *Plugin ginger :*
> System
> Software Update
> Repositories
> Restart, Shutdown of Host
> Hot-plug
> Networks
> Storage
> Security
> Firewalls
> SELinux
> PAM Config
> Configuration
> Dumps
> Backups
> Systemd
> ...
> Console
> Users and Groups
>
> *Plugin ginger-ppc :*
> Firmware Updates
> IBM SEP
> Energy Management
Energy management isn't exclusive to Power. It works in x86_64 too.
>
> *Plugin ginger-s390x :*
> HPM
> Performance Metrics and Diagnostics
This feature can be multi-architecture if the tools used to get the
metrics/diagnostics aren't
exclusive to Z.
Makes sense, we should try to make this one platform independent. I
guess we are not there yet, but we should plan for.
> zIO Management
>
> ***Note* : Hot-plug, Security, dumps and systemd configurations etc
> will be kind of future functionality that can be part of base ginger
> plugin.
>
> *Key points to consider:*
> 1. At the moment separation of the ginger-ppc part of functionality
> (mentioned above) from the ginger may not be the highest priority.
> 2. Shall we keep Users and Groups part of ginger plugin or do we want
> this be part of base WOK frame work ?
The idea is to keep the WOK framework 'featureless'. It will consist
only in the cherrypy server,
model/control APIs and a 'Welcome to WOK' message when no plug-in is
installed.
Keeping wok featurelss makes lot of sense. I guess what we need to think
about is the scope of user and roles management. From what I currently
see there are currently 3 roles/profiles defined: Administrator, Kimchi
User and Virt User. With the agreed feature split User Management is
available only if Ginger is installed, which is true with the current
stable implementation as well. Since in the current implementation
Kimchi is always available we don't have an issue offering Kimchi and
Virt User profiles by default, but after the separation these profiles
make sense only if Kimchi plugin is installed. Theoretically any new
plugin could bring in the requirement of new profiles.
The other question we have to clarify is: do we need User Management
when Ginger is not available as plugin ?
From the implementation I understand that we are talking here only
about locally (on this host) defined users and roles, ldap user
management isn't performed. From this point of view it's fitting in the
scope of Host Management. The part not really fitting is the plugin
specific profile aspect.
Since we want to keep the plugins as independent as possible from each
other, one option I see is: if a certain plugin has the requirement for
a certain user role it has to expose the description of this role in a
tbd. format to the User Management feature.
>
> Let me know what do you think about this !!!
Just a reminder: we still need to figure it out how we'll change the
current engine to support a
plug-in extending an existing tab from another plug-in.
We are working on that one.
If it proofs to be too much of work or
uggly hack, I would propose to leave it how it is for now in the old
UI-design (meaning keep Administration tab) and do the final
UI-restructuring based on the new UI-Design.
And another reminder: Ginger mailing list is now
ginger-dev-list(a)nongnu.org. Feel free to join
the ML:
https://lists.nongnu.org/mailman/listinfo/ginger-dev-list
Daniel
>
> Regards
> Chandra
> On 08/07/2015 09:51 PM, Daniel Henrique Barboza wrote:
>>
>>
>> On 08/07/2015 01:14 PM, Walter Niklaus wrote:
>>>
>>>
>>> On 07.08.2015 18:05, Aline Manera wrote:
>>>>
>>>> Seems we are close to a conclusion on that subject. ;-)
>>>>
>>>> Let me resume what we have already agreed:
>>>>
>>>> 1) Move software update + repositories + restart + shutdown +
>>>> console to Ginger
>>>>
>>>> 2) Create a new plugin for host basic info + host stats + debug
>>>> reports
>>>>
>>>> My proposal is to have this new plugin under Ginger community
>>>> responsibility.
>>>> AFAIU, the Ginger community is for a host management plugin. And
>>>> as this new plugin is for basic host details, it makes all sense
>>>> for me.
>>>>
>>>> Thinking on that, I suggest to name this new plugin as
>>>> 'ginger-basic' which will load a tab named 'Host' with
host basic
>>>> information + host stats + debug reports.
>>>> And 'ginger' plugin will extend this Host tab generated by
>>>> ginger-basic.
>>>>
>>>> It will increase the Ginger community scope and make everyone
>>>> happy! =)
>>>> We can also think about splitting ginger into: ginger, ginger-ppc,
>>>> ginger-z, etc... (but it is an other different topic to be
>>>> discussed on Ginger community)
>>>>
>>>> So from a user and devel perspective, the whole Kimchi Host tab
>>>> will be moved to Ginger community into 2 different plugins:
>>>> ginger-basic and ginger.
>>>> In a simple phrase, Ginger plugins provide a Host tab based on wok
>>>> framework.
>>>>
>>>> To do not get the user crazy "Where is Ginger now?" while
loading
>>>> the updated version, I have 2 suggestion:
>>>>
>>>> A) Work with plugin logos!
>>>> Maybe a pan design to represent wok framework and on top of it
>>>> all the plugins logos are displayed.
>>>> To do that we will need a logo for wok and ginger.
>>>>
>>>> B) Create a introduction page to show the differences from the
>>>> older version.
>>>> The idea behind it is like some Android applications do when
>>>> get updated. Once you open it, a page is displayed as a layer on
>>>> top of the application with balloons and arrows to tell "this
>>>> feature was moved to here"...
>>>>
>>> I'm ok with both solutions although since we are planning to
>>> redesign the UI in order to improve the user experience I hope that
>>> we are able to do a very good job and make the new UI-design
>>> intuitive enough for the user to find the right menue points :-)
>>>
>>>> Do you agree, Daniel and Walter?
>>>>
>>> I agree.
>>
>> I agree. Unless someone else has a strong point against it, we can
>> move forward with this idea,
>> starting discussions about what needs to be done and how long we'll
>> need to make it happen. From
>> the top of my head the most critical point is how to make the
>> current engine support a plug-in
>> extends an existing plug-in. We can start there.
>>
>>>
>>>> Regards,
>>>> Aline Manera
>>>
>>
>> _______________________________________________
>> Kimchi-devel mailing list
>> Kimchi-devel(a)ovirt.org
>>
http://lists.ovirt.org/mailman/listinfo/kimchi-devel
>>
>
_______________________________________________
Kimchi-devel mailing list
Kimchi-devel(a)ovirt.org
http://lists.ovirt.org/mailman/listinfo/kimchi-devel