[sos-devel] [networking] add ethtool -d in networking plugin

[Sheng Liu]

Hi Bryn:
the user wanted to confirm some bug .Capturing the grcDump using the
'ethtoold -d' command would allow them to make that confirmation.
Another method to capture the grcDump is through the NetXtreme II eDiag
Diagnostic Utility.  However, this utility could not to be used by
customers.

On 07/09/2014 07:19 PM, Bryn M. Reeves wrote:
> On Wed, Jul 09, 2014 at 04:51:10PM +0800, Sheng Liu wrote:
>> Subject: [networking] add ethtool -d  in networking plugin
>> Date: Wed, 02 Jul 2014 11:11:14 +0800
>> From: Sheng Liu <shenglsh at linux.vnet.ibm.com>
>> To: sos-devel-request at redhat.com
>>
>> add ethtool -d <eth> command to networking plugin to collect more info.
>>
>> Signed-off-by: Sheng Liu <shenglsh at linux.vnet.ibm.com>
>
> Thanks for the submission. Can you explain a bit more why this is
> valuable enough to collect by default?
>
> We've had some recent experience with the less common ethtool interfaces
> that makes me think this may present a risk that outweighs the likely
> rare occasions that it is useful for debugging.
>
> A bug in the qlcnic ETHTOOL_GSTRINGS ioctl causes severe memory
> corruption during 'ethtool -S' on affected hardware. It's fixed upstream
> but the possibility of this type of problem (which is more common in the
> less frequently exercised ioctls) means we need very good justification
> to add this to the default run for all users.
>
> I'll canvas around the network engineers that I know to see how
> frequently this information might be useful but can you provide a more
> detailed explanation of why you think it's valuable? If nothing else
> for change like this that rationale should be included in the commit
> message.
>
> Regards,
> Bryn.
>