On Mon, 2011-01-24 at 15:00 +0100, Stefan Richter wrote: > On Jan 24 Rudy Zijlstra wrote: > > On 01/24/2011 01:21 AM, James Bottomley wrote: > > > On Sun, 2011-01-23 at 11:41 +0100, Rudy Zijlstra wrote: > > > > > >> Dears, > > >> > > >> I got the following crash by: > > >> - cp several G onto a raid 5 on a Marvell 88SE6480//// based controller > > >> while also doing > > >> - cp * ../test/ on the same raid. > > >> > > >> I strongly suspect mvsas to be the cause > > >> > > > I've got to ask why? > > > > > cause the 88SE6480 has been giving grief for a long time, and i was > > loading that one specifically. > > you may well be right (i do not read crash info i fear), but a bit > > strange to have NFS crash when nfs was not much used at that moment > > > [...] > > >> [ 2821.393697] Pid: 1105, comm: rpc.mountd Not tainted 2.6.37 #1 > > >> X7SB4/E/X7SB4/E > > >> [ 2821.393697] RIP: 0010:[<ffffffff815a1b5c>] [<ffffffff815a1b5c>] > > >> cache_revisit_request+0xab/0x105 > > >> > > > This says the bad deref occurred in the sunrpc authentication cache. > > > Nothing at all in the trace implicates mvsas ... in fact nothing even > > > remotely relates to it at all. It really looks like an NFS problem. > > Perhaps a silent memory corruption, with an innocent bystander becoming a > victim? > > Rudy, if possible repeat the test with NFS completely shut down and > disabled. I'll schedule that for Saturday morning. Earliest time i can disable NFS without user impact and do another stress test on mvsas. And i need to be physically close to the system, as a kernel crash needs a reset button to recover. -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
