[ofa-general] Re: [PATCH 0/4] opensm: Unicast Routing Cache
Sasha Khapyorsky
sashak at voltaire.com
Sun Jun 29 14:50:13 PDT 2008
On 00:33 Mon 30 Jun , Sasha Khapyorsky wrote:
>
> Another thing we may want to add (but it is not related to routing cache
> or incremental routing) it to keep two sets of LFTs with switch object
> for validation purposes - "requested" (filled by routing algorithm) and
> "real" (filled from responded MADs).
Actually even this is not needed. If OpenSM tried to set some LFT block
and failed it is better to try again (or re-request a heavy sweep), such
validation could be done on per-block basic in LFT MADs receiver -
nothing fancy.
Sasha
More information about the general
mailing list