Hi Lars,

First, thanks very much for considering OpenHPI as a reference for the
test suite. I frequently use hpitest when preparing a release. We
certainly need the ability to fully test the FUMI interface in automated
regression tests.

I suggest that your efforts would be much more valuable in enhancing the
existing simulator plug-in instead of modifying the ipmidirect plug-in
for testing purposes.

I agree that it would be useful to snapshot an existing real world
hardware resource configuration of a system. However, the user would be
limited to only the HPI capabilities provided by that specific system.
Instead, if the simulator plug-in were modified to read its resource
configuration from a text file, any possible resource configuration
could be tested - even those that are not implemented on any existing
device. A snapshot of an existing system generated by your new client
could provide a quick starting point for a custom configuration. The
simulator plug-in currently has its resource configuration hard coded in
arrays within the file plugins/simulator/sim_resources.c.

For simulator controls, the routines in plugins/simulator/sim_injector.c
could be useful. For instance sim_inject_event() could be used to
generate resource failure and hotswap events, as well as FUMI events.

I don't think the ipmidirect plug-in currently supports any of the FUMI
APIs. In fact, I don't think any of our current plug-ins set
SAHPI_CAPABILITY_FUMI.

In my opinion, it would probably be easier to enhance the simulator
plug-in than to replace the back end of the ipmidirect plug-in and add a
simulation control interface.

Best Regards,
Ric White


On Wed, 2009-12-09 at 16:16 +0000, Lars Wetzel wrote:
> Hi
> 
> I want to present you some ideas for a new simulator plugin and invite you to 
> comment it or start a discussion.
> 
> Motivation/Background: 
> Within the SAF hpitest workgroup we start to implement test cases for FUMI. 
> At 
> the moment we have the problem to verify our test cases due to a missing 
> "default" FUMI implementation. Especially with FUMI it will be necessary to 
> interact with a hpi-b daemon to run regression test cases and maybe the one 
> or other functional test case (e.g. load an upgrade file, start an upgrade). 
> So we had the idea to use the openhpid simulator as reference for the test 
> suite. 
> 
> Ideas:
> The existing simulator plugin doesn't cover the requirements from above and 
> so 
> I start to implement a new simulator plugin with following edge points / 
> ideas:
> -- With a new hpi-b client "eg. hpigensimdata" a snapshot is made of an 
> existing system by generating a text file with hpi-b information about the 
> running system.
> -- This file will be parsed at the start of the new simulator plugin.
> -- The same interface (text file) will be used to work as 'upgrade' file: 
> Restrict the data to one resource and/or change some rdr data.
> -- For this purpose I thought it could make sense to adapt the ipmidirect 
> plugin: 
> ---- Replace the lan/smi connection by parsing the file
> ---- Instead of generation IPMI commands work with the cache and answer 
> properly
> ---- Introduce "simulator" controls: E.g. mark a resource as failed. 
> 
> I'm really at the start, which means I have some code but nothing which can 
> be 
> compiled or is ready for any publishing process. If someone has interest to 
> work with me, I will share the code. It's really open in which direction this 
> could go ... Also if you mean it makes no sense, I will stop it.
> 
> Thanks
>       Lars
> 
>   
> 
> ------------------------------------------------------------------------------
> Return on Information:
> Google Enterprise Search pays you back
> Get the facts.
> http://p.sf.net/sfu/google-dev2dev
> _______________________________________________
> Openhpi-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/openhpi-devel


------------------------------------------------------------------------------
Return on Information:
Google Enterprise Search pays you back
Get the facts.
http://p.sf.net/sfu/google-dev2dev
_______________________________________________
Openhpi-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openhpi-devel

Reply via email to