Hi, Finally I had time to give a new shiny kernel a shot on my machines and unfortunately sparc didn't survive ;) So ... let me advise you to actually _have_ backup. The story is short: I crosscompiled new kernel (f3b8436a), installed it over nfs, rebooted into the new kernel and at that point everything looked fine. I run gentoo on sparc so I started emerge system and emerge world (heavy I/O stuff) etc. My sparc is not a spead deamon (2x450MHz) so I left home and a few hours later the box was more or less dead or more precisely I/O to disk and nfs share was dead. In dmesg I saw info about stale nfs handle, ext3 corruption, access beyond end of device, some trace with scsi_* in it, etc. Interesting fact is that somewhere at that time my switch decided to stop switching - so that's why NFS stopped first (I guess). Unfortunately I didn't capture dmesg output as I thought I could bisect that and provide some more info. Other architectures (x86_64 and powerpc) I run this kernel on now are fine. They have similar configuration (some shares over nfs + ext3 fs). The obvious difference is sparc uses scsi. So this may be not related to NFS at all. Don't know - just guessing. I ended up with corrupted ext3 (see sparc.jpg) and working now to restore it. You've been warned 8-) Mariusz
Attachment:
sparc.jpg
Description: JPEG image
Attachment:
config.gz
Description: GNU Zip compressed data
