Update on fs-corrupting kernel bug?

[robert.rankin.jr]

Based on responses in the bug report and my own experience, I'm pretty sure it's a kernel issue. Maybe the nVidia driver screws this kernel up. I'm no expert, so I can't comment. I also have no clue about the -13, but I've heard that the -14 (in jaunty-proposed) may fix the problem. I'm still using the 32-bit version because trying to operate with a 64bit installation is too difficult. If you want to use Jaunty, I'd use the 32bit version. It is confirmed, though, that you can install the 64bit version and upgrade the kernel. Check out this:

https://bugs.launchpad.net/ubuntu/+s...1/comments/122

You don't need to do the intrepid install, you can just follow the instructions from within a working Jaunty installation on the first boot. Or, if you're paranoid like me, and neither like upgrades nor screwed up installations, you can install from the usb-disk, reboot, chroot into the new environment and do the kernel install.

Robert

[gila_monster]

Good info, Robert. Thanks.

As promised, and in case anyone wants to try it, the instructions for the new nVidia driver are here:

http://ubuntuforums.org/showthread.php?t=1125400&highlight=nvidia+185+installation

I had no trouble with this, but I've done this kind of thing before. If you are not sure of what you are doing, I recommend not trying this unless you are looking to learn some things. When it goes well, it's wonderful. When it doesn't, it can take a while to clean up the mess. The instructions are clear, but a bit complicated.

[glacialfury]

For those wondering about the filesystem mattering, I've had this occur on a serp4, Ubuntu and Kubuntu 64 bit jaunty under both ext3 and ext4. I haven't looked into nvidia involvement yet though.

[thomasaaron]

glacialfury, I don't think there is an nvidia element to it. I believe it has to do with some change in the SATA controller drivers.

[phyzome]

glacialfury, I don't think there is an nvidia element to it. I believe it has to do with some change in the SATA controller drivers.

Have you been able to reproduce it on a test machine?

[thomasaaron]

Working on it right now. I have a SerP4 on my desk and I'm doing some huge file transfers with it. Hopefully that will crack it.

JDB says he thinks it doesn't happen on a 32-bit kernel. Problem is, that *might* forfeit suspend functionality.

[glacialfury]

I wonder. Most of the relevant bug reports point toward it being a specific kernel; x.x.x.x-11, whichever one ships with the 9.04 installation, and affecting primarily (exclusively?) the 64-bit version. It may not be worth your while to hunt down the hardware portion of it specifically, since it seems it is triggered by the kernel, and it can be fixed relatively easily by changing the kernel.

I followed a two-line command to download and swap out the kernel, and it hasn't had any problems since then.

[thomasaaron]

Good points...

Could you please post your process?

[samalex]

Just curious I have a Pangolin Performance in transit now. Will it be preloaded with the 2.6.28-11 kernel? If so is it recommended to upgrade to -13 or whatever the latest is in the Ubuntu repositories?

Just checking, I don't want to drop all my stuff on the laptop then run into file system corruption.

Thanks ...

Sam

[thomasaaron]

samalex,

I *think* we're shipping with the *-13 kernel, and the problem does seem related to large file transfers.

However, it's not clear to me yet if this problem is affecting the PanP5 -- but it may be. I'm running a number of tests right now.

This bug is a priority, and we're wrestling with it quite actively.