Ubuntu
linux package

  1. Bug #285892
  2. Comment #97

Comment 97 for bug 285892

Revision history for this message
Raj B (bigwoof) wrote :

fwiw, I am having the same problems with the latest Natty packages.

The kernel is
Linux mythtv 2.6.37-11-generic #25-Ubuntu SMP Tue Dec 21 23:42:56 UTC 2010 x86_64 GNU/Linux

My setup is
Asus P7P55D-E LX Motherboard with the Marvell 88SE9125 SATA 3 controller
Intel i5 760 CPU
PC1333 8G RAM (4GB x 2 Corsair)
5 Sata 2 2 TB Disks

1 Sata 3 Seagate 2 TB Disk (used as the boot disk and the disk on which everything below is being written). plugged in as SATA 3 using AHCI

1 PVR 150 Video Capture card
Asus ENGT430 Graphics Card

I'm getting a ton of
[ 5884.881538] ata9.00: failed command: WRITE FPDMA QUEUED
[ 5884.909807] ata9.00: cmd 61/08:20:f0:71:c5/00:00:73:00:00/40 tag 4 ncq 4096 out
[ 5884.909810] res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
[ 5885.022592] ata9.00: status: { DRDY }
[ 5885.051123] ata9: hard resetting link
[ 5885.591937] ata9: SATA link up 6.0 Gbps (SStatus 133 SControl 370)
[ 5885.593771] ata9.00: configured for UDMA/133
[ 5885.593996] ata9.00: device reported invalid CHS sector 0
[ 5885.594002] ata9.00: device reported invalid CHS sector 0
[ 5885.594013] ata9: EH complete

errors in the logs. I noticed this because ivtv was outputting a ton of "Unable to Save MPG stream" errors. I thought it was a bug in ivtv but now I realize that it was a SATA 3 error and that the drive had become read-only.

this has happened a few times now and the system is locked up hard each time. It can ping but nothing is running. I had ssh access during one of these events and nothing worked (reboot, all process were in zombie state, etc.). which makes sense as the root drive was now read-only and inoperable.

I've lost data because of this as well. my entire /var/lib/mysql directory was blown away and recovered into lost+found. other directories are there as well.

I'm going to a) switch the sata 3 drive to the sata 2 controller, and b) reinstall ubuntu (as I'm not sure what went missing with the latest crash). I'm a little surprised that this bug has remained through multiple kernel revisions.

Raj