Not sure if this is relevant, but we've been running a 52-disk Linux software RAID10 array since January with no isssues re: deadlocking. I patched our kernel to use seq_file for /proc/mdstat, but other than that, I'm using a 'standard' Red Hat kernel. I cannot comment on ENBD as I have never used it. I may be an exception and not the rule, but I haven't heard of anyone complaining about RAID on RAID on this list and from my personal experience it has been stable. Regards, Andy. On Mon, 2003-07-14 at 07:21, Bernd Schubert wrote: > On Friday 11 July 2003 16:57, Peter T. Breuer wrote: > > "A month of sundays ago Ulrich Hahn wrote:" > > > > > scheme-1 and scheme-2 with software-raid on the fall-back-server will > > > not work out. I'm not sure if sw-raid over sw-raid is prone to deadlocks > > > (Peter?), but the tests I did showed that extremly low rebuild-rates, > > > even > > > > It's impossible to say. The raid code is sophisticated = obscure. > > I think there are no deadlocks - it maintains a separate local cache > > of request structs per device, as I recall. Ummmm ... but what about > > buffers? It may maybe deadlock for buffers if there is enough copying > > and it's done synchronously ... > > > > > with tuned md-parameters, will make any medium sized raid unusable. > > > > Thanks Ulrich and Peter for your answers and help, that helps me a lot for our > future plans. > > Best regards, > Bernd > - > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- Andrew Rechenberg <arechenberg@shermfin.com> Infrastructure Team, Sherman Financial Group - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
