On Wed, 2011-02-09 at 15:54 +0100, Martin Husemann wrote: > On Wed, Feb 09, 2011 at 11:02:06PM +1100, matthew green wrote: > > wild guess: can you re-seat the cpu modules? or the ram? or perhaps > > try increasing the timeout from 1000 to 10000? (see ipifuncs.c.) > > FWIW, I have seen this happen on my SB1000 as well, only under realy > heavy load, and in my case not quite reproducable. > > I don't see how this could happen though. Maybe we should scale the DELAY() > slightly with bigger loop counts? > > Martin > I've seen that the delay is called with a fixed value of 1. May be a slightly bigger random-value would help here, too. - Volkmar
Attachment:
signature.asc
Description: This is a digitally signed message part