[ofa-general] RE: [Bug 465] IPoIB CM HA fails after several hours of failures

[Hal Rosenstock]

On Thu, 2007-03-29 at 01:08, Michael S. Tsirkin wrote:
> > > > > I expect ibnetdiscover can do this, but was unable to grok
> > > > > the output syntax. 
> > > > 
> > > > I'll explain once I have the answer to the above question.
> > 
> > Search for "H-<GUID in hex>", where GUID in hex is the node GUID, in the
> > output of ibnetdiscover. [1] to the right of it indicates it is port 1.
> > 
> > So for example, 
> > Switch  24 "S-005442ba00003080"         # "ISR9024 Voltaire" base port 0 lid 6 lmc 0
> > [22]    "H-0008f10403961354"[1]         # "MT23108 InfiniHost Mellanox Technologies" lid 4
> > 
> > It is listed under the switch it is attached to and in the right hand
> > side is the LID of the switch which in this case is 6.
> 
> And how do I know the switch port here?

It's the port on the left hand side (e.g. 22 in this case).

> Hal, where does this syntax come from? Some legacy script?

It originated with a proprietary tool a long time ago and has been
enhanced from there to add additional information as time went on.

> How about fixing this tool to provide a sane,

I think it is sane output.

>  tabulated output, with
> a top self-documenting header, and flags to select specific rows/colums?
> 
> I envision something a la ps:
> 
> type guid port remote_lid remote_port description

I'd like to understand more of what you intend and why the current
format is insufficient.

> Would such a patch be accepted?

Such a patch would need to include all the affected tools, not just
ibnetdiscover or alternatively have an option for this new output format
with the existing format as a default. I think the latter would be
needed for backwards compatibility.

-- Hal