On Fri, 2013-06-28 at 01:28 +0200, Martin Braure de Calignon wrote: > Le jeudi 27 juin 2013 à 18:31 +0100, Ben Hutchings a écrit : > > On Thu, Jun 27, 2013 at 07:14:57PM +0200, Martin Braure de Calignon wrote: > > [...] > > > So you think that this temperature stuff could be related to my ext4 > > > problems? > > [...] > > > Are you overclocking the processor? > First of all, thank you for your support. > > Well I was unsure, so I opened the computer, removed the CPU fan. > Processor is a 1.7Ghz, but in the BIOS the CPU clock was 110 where it > should be 100 to be 1.7Ghz. > So yes it was. It's no longer. > Second problem, my CPU fan was controlled by a potentiometer which was > not turned to it maximum. I just removed it and connected the cpu fan > directly to the mainboard. I'm not sure of that, but I'm not sure there > was enough thermal paste between the processor and the fan. So it is > still a future track if the problem reappeared. > Third problem the PC was full of dust. I just cleaned it. > Fourth problem [1], it seems (on purpose in linux source code) that my > processor is not happy with cpufreq governance to ondemand and was > automatically switched to performance.
Right, I guess this is an old Pentium 4 which can't change frequency very quickly. [...] > I just rebooted with the new CPU frequencies, and tried everything I > copy a big directory from my (new) ext3 partition to the raid1 ext4 > device (which currently now has only 1 device in it ;)) > no problem during like 15 minutes... > So I tried something else, because I'm not sure, but I've noticed that > problem occurred more when I'm deleting files, so I run a fdupes on > another big directory of that raid1 ext4 device. > and boum [2]! [...] > [2]: > [...] // lines before are just boot of the system, then the manual mount > of the raid devices and the other devices. I ran a fsck.ext4 forced > right before. > > [ 1618.164009] EXT4-fs (md0): mounted filesystem with journalled data > mode. Opts: data=journal > [ 3341.908050] EXT4-fs error (device md0): ext4_lookup:1050: inode > #54003553: comm fdupes: deleted inode referenced: 54003610 > [ 3342.327029] EXT4-fs error (device md0): ext4_lookup:1050: inode > #54003553: comm fdupes: deleted inode referenced: 54003610 > [ 3342.456047] EXT4-fs error (device md0): ext4_lookup:1050: inode > #54003553: comm fdupes: deleted inode referenced: 54003604 [...] > any clue? :/ I still suspect a hardware problem, just because ext4 is the default filesystem for 'wheezy' and no-one else reported this yet. > Few other comments: I have all other FS under ext4 (but the new ext3 i > created this afternoon) and never got problem with them. > My PC originally had only IDE BUS, so I added two PCI controller for > having SATA. > First two disks on the first controller (a big LVM) never got a problem > with ext4 > the other two disks on the second controller (different brand) is the > RAID1 device. So maybe the second controller (or its driver) is faulty. Can you try using the RAID-1 disks connected to the first controller, with nothing connected to the second controller? Or do you need the first two disks in order to boot? Ben. -- Ben Hutchings Knowledge is power. France is bacon.
signature.asc
Description: This is a digitally signed message part
