From: Justin P. <jp...@lu...> - 2008-08-30 21:14:07
|
On Sat, 30 Aug 2008, Jonas Petersson wrote: > Hi Justin and thanks for your prompt response, > > That would be /sys/block/sda/device/queue_depth, right? I already checked > that and it was "1" so I guess there is nothing to do. Correct.. >> Also, do/not run smart tests during heavy I/O? > > They are certainly OK without heavy I/O. Those that take some time to run > where run during "normal" I/O - I kept using the system normally. The problem > occurs during low I/O conditions as far as I can tell. For example during the > night (slocate may run at some stage, but according to the logs the problems > has occured at 4 different time during the night and even a fairly long > slocate session can't possibly have covered more than 2 of them at worst). I do not have the same exact problem you have, but I have a similar problem: [189770.219773] ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 frozen [189770.219784] ata1.00: cmd 35/00:40:9a:d9:7a/00:00:12:00:00/e0 tag 0 dma 32768 out [189770.219786] res 40/00:ff:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout) [189770.219790] ata1.00: status: { DRDY } [189770.219795] ata1: hard resetting link [189770.524770] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300) [189770.543960] ata1.00: configured for UDMA/133 [189770.543977] ata1: EH complete I think its a chipset issue personally, I've spoken to various people about it on LKML, and other mailing lists basically there is no ultimate box, I've tried disabling noapic, apci=off etc, no luck. I just ignore it, it doesn't seem to affect anything (in my case) as of yet and via smart tests/etc I cannot/do not see any form of drive failure, all tests pass normally and the smart output looks good. > > >> Have you tried disabline smartmontools to see if the problem persists? > > Yes: Initially, it happened before I even installed smartmontools. It kept > happening with while smartd was running for a while. It kept happening when I > disabled smartd and just ran smartctl now and then. It also happened (one) > about 5 minutes after rebooting when no smartmontools programs either smartd > nor smartctl had efter run. In this particular case, the boot process > involved automatic installation of the nvidia binary driver which may have > caised a bit of heavy I/O for a while. > > Does this give you any further clues? Would a full dump of smartctl --all > help? smartctl -a would be useful (#1) An "lshw" run would also be useful, I am curious which chipset they are using for the SATA I/O on that hardware type. Justin. |