Update on fs-corrupting kernel bug?

[samalex]

Hi Thomas,

Awesome, thanks. When the laptop arrives (hopefully Friday) I'll check the kernel version and run it through the ringer before transferring any huge amounts of data to it. If you guys do find a fix or have suggestions on updates, I'll keep an eye on the forum.

Thanks --

Sam

[phyzome]

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

This is good to hear.

Let me know if I can help. I have a spare 6GB partition on my panp4i, and I'd be willing to install with a couple different kernels.

[thomasaaron]

https://bugs.launchpad.net/ubuntu/+s...1?comments=all

That's the pertinent launchpad report. There are a couple ways to fix it, for sure. But I don't really recommend either just yet.

One could install 32-bit Ubuntu. But who wants that?
There's also a round-about fix on post 122 or 123. Not ideal.

The devs are under HUGE pressure with this one, and it's fixed in Karmic -- and they've promised we won't have to wait for Karmic for the fix. It's largely a waiting game at this point.

Until I find out more, I'd minimize large file transfers.

[gila_monster]

I did try the solution in post 122, and it seemed to work, but I did not hammer the system particularly hard after that.

Currently running the 2.6.18-14 kernel, and no issues, but again not hammering hard.

For fun, I'm attempting to install Gentoo on the spare (Gateway) laptop. Have had zero success...and yet, it's less frustrating than the current problem!

[thomasaaron]

Thanks, gila_monster.

OK, I've got an SerP4 computer in my shop. This is one that a customer sent in for diagnosis. It has been hit by this bug.

Right now, I'm running the *-14 kernel. I've downloaded almost 30 GB of data to it today, and I'm not seeing any problem at all.

fsck keeps giving it a clean bill of health.

Is there anyone out there having this problem on the *-14 kernel?

There was a weak hint in that bug report that it *might/maybe/fingers-crossed* be fixed in *-14.

[glacialfury]

I hate to revive this dreaded topic, but I think I might have this in Karmic-alpha5. I did a clean install of Karmic about two weeks ago, which initially worked fine, and then gradually started having errors, culminating in the "Can't boot, run fsck" sequence we all know and love. Since then I've simply been using my Windows partition for my work, but I'm sure you'd all like me to return to my happier opensource self.

The information about this bug is not well delineated and I don't have the time to experiment that I used to; what can I do to ...
a) confirm whether this is, in fact, the same bug
b) repair or create a workaround

I'm happy to run any commands or output for you lot, and there's nothing critical on the damaged partition, so we can do whatever you like to it.

It's a SERP4, plagued by problems in nearly every version.
8.04-8.10: kernel oops related to clocksource causes block
9.04-9.10: filesystem corruption caused by ?kernel problem

I'm also willing to try another distro if you feel this will alleviate the problem. I like Ubuntu very much, but exploring is also fun, if I can expect stability out of something else.

Regards,

[thomasaaron]

It's difficult right now to tell if it is the same bug. Did Karmic install with ext4 by default? If so, there could be a problem there too. We've not started testing with Karmic yet, and probably won't for a couple of weeks.

Also, I've heard it doesn't exist in 32-bit.

[jdb]

It's difficult right now to tell if it is the same bug. Did Karmic install with ext4 by default? If so, there could be a problem there too. We've not started testing with Karmic yet, and probably won't for a couple of weeks.

Also, I've heard it doesn't exist in 32-bit.

WOW! if it's not available in 32 bit there is a lot of hardware out there (here) that will be left behind.
Out of 7 boxes in this house, only 3 are 64 bit machines.

jdb

[phyzome]

WOW! if it's not available in 32 bit there is a lot of hardware out there (here) that will be left behind.
Out of 7 boxes in this house, only 3 are 64 bit machines.

No, the *bug* isn't available in 32-bit.

[HotShotDJ]

Thomasaaron --

I've gone through the above linked bug report regarding this issue in preparation for the imminent arrival of my PanP5. I have a thought regarding this problem until Karmic goes gold.

According to the bug report, the problem is also fixed in the 2.6.30 kernel. I've been using that kernel as a drop-in replacement on my current system (32 bit) since its RC days without an issue due to the Intel Video problems in Jaunty. It is a pretty painless install - here is the code for getting and installing the 64 bit version:

$ wget -c http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.30.5/linux-headers-2.6.30-02063005-generic_2.6.30-02063005_amd64.deb http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.30.5/linux-headers-2.6.30-02063005_2.6.30-02063005_all.deb http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.30.5/linux-image-2.6.30-02063005-generic_2.6.30-02063005_amd64.deb
$ sudo dpkg -i linux-headers-2.6.30-02063005-generic_2.6.30-02063005_amd64.deb linux-headers-2.6.30-02063005_2.6.30-02063005_all.deb linux-image-2.6.30-02063005-generic_2.6.30-02063005_amd64.deb

On the down side, this kernel will break AppArmor, so people who need AppArmor won't be able to use it. I don't know if there are other problems that I didn't think of. As I said, I've been using it on a 32 bit Dell system, so I don't know what unintended consequences the 64 bit 2.6.30 kernel might have on a System76 computer.