Re: Re: Proposal for new component in Tizen IVI

Wed, 16 Apr 2014 04:57:48 -0700

Hi Jussi,

 

Thanks for your valuable inputs. Please find my comments below inline for your queries.

 

Regards

Abhishek

 

------- Original Message -------

Sender : Juha-Matti Liukkonen<[email protected]>

Date : Apr 16, 2014 01:32 (GMT+09:00)

Title : Re: Proposal for new component in Tizen IVI

 

Hello Abhishek and others,

considering that HM essentially tracks systemd notifications and stores them into a DB, my initial impression also is that this would be a logical extension of systemd itself. Or possibly achieved using the Exec[Start,Stop][Pre,Post] directives, although that would not be as transparent.
<Abhishek> From the modularity and ownership prespective it is better to have this functionality seperate from systemd.  Besides as Systemd is an independent upstream project used by many distros, it may be difficult to push tizen IVI specific changes there.
 
The concept of definable recovery strategies is interesting. Could you elaborate more on the HM Recovery Client, and how it would interact with systemd and murphy?
<Abhishek> Recovery module shall be a OEM specific module, where they can implement their own recovery strategy like checkpointing/restart, Function Inhibition, resource management etc. For Audio, video and network specific scenarios Murphy could address these functionalities. We could also envision that HM for those specific aspects interacts with Murphy. However, there are many other system aspects which are outside the purview of Murphy and those could  be manged from the proposed HM when it is fully realized.

Best regards,
- Jussi Liukkonen

On 15.4.2014, at 18:21, Stoppa, Igor <[email protected]> wrote:

Hello Abhishek,

please find few comments below.

Comment 1:
maybe I'm being tricked by the google document viewer, but the
presentation seems to lack the last point:   11 Use Cases

Comment 2:
I could understand why the low level part of the functionality
described (process watchdog and stats) could not be merged into
systemd.
iirc systemd already does a lot wrt process WD, so it would seem
natural to enhance systemd with more extensive logging capability.

Comment 3:
most of the higher level logic could be implemented using Murphy,
which is already present on IVI platform, afaik.
But there is no mention of Murphy and why it would not be suitable
for the job.


To summarize:
enhancing & reusing existing components would seem preferable
to introducing yet another one, whenever possible.

Maybe it's not possible and for very good reasons, but I couldn't
figure out why, by reading the presentation.

I think it could be enhanced with a more detailed description of the
problem, by adding real life use cases, for example.

And a brief analysis of existing solutions and explanation of why they
are not suitable would also put the reader in a better mindset,
to understand why a new component is needed.

cheers, igor


On 15 April 2014 08:30, Abhishek Sharma <[email protected]> wrote:

Dear All,


 

We would like to propose a new component called System Health Manager for Tizen IVI.  It is imperative that automotive platform should provide a means to monitor critical services and take corrective action whenever a failure or deviation from expected services is detected. In that respect, the proposed Health Manager (HM) shall seamlessly integrate with Tizen IVI for providing a limited fault-tolerance capability. Attached please find the presentation slides which describe overall functioning and features of the proposed HM. We request you to provide your comments/feedback on our proposal, and ways for taking this initiative forward.


 

The key features of the proposed HM include detecting disruption of services, logging data required for diagnosis and applying directed recovery strategies.The proposed HM uses systemd interfaces and signals to monitor configured critical services and recover them, if failed.

 

The initial development is undergoing and can be released at tizen.org for open source contributions following the feedback and internal due process for releasing the code. At the moment, we have basic structure for HM core, Dbus Communication and Fault-database. We have also designed a placeholder for dummy recovery client at the moment. Going forward this needs to be developed as per system requirements.


 

Looking forward to your suggestions, feedback and active contributions of this feature on tizen.org for IVI profile.


 

Regards

Abhishek


 

 

<201404151100980_QKNMBDIF.gif>


_______________________________________________
IVI mailing list
[email protected]
https://lists.tizen.org/listinfo/ivi




--
cheers, igor
_______________________________________________
IVI mailing list
[email protected]
https://lists.tizen.org/listinfo/ivi

-- 
Juha-Matti Liukkonen
Technology Director
Reaktor / Embedded Software
+358 40 5280142

 

 

 

_______________________________________________
IVI mailing list
[email protected]
https://lists.tizen.org/listinfo/ivi

Reply via email to