Misbehaving Ubuntu box, normally well behaved and reliable, needs help.

April 4, 2008 9:42 AM

UbuntuFilter:Linux people - how would you troubleshoot this one? All of a sudden my sound card doesn't exist, my CD burner doesn't burn, and fusermount doesn't want to work anymore. Everything was fine on Monday. Any suggestions on how to figure out WTF is going on?

The longer story:
My work desktop is running Ubuntu Dapper. I came to work on Tuesday to find I had left my streaming audio on, ie I was still hearing audio through my headphones. However - my box was in an odd state. The keyboard was not responding, the mouse was responding sporadically. I couldn't log in - no keyboard. I tried to ssh into the box from another machine, but that didn't work - I didn't get any response. At that point I just rebooted via the power button. While the box rebooted, fsck was run - it apparently found and fixed some errors in the root partition, and rebooted itself. Finally, once the box was finally up and I was logged in, I noticed the soundcard (a crappy onboard soundcard) was apparently no longer installed.

I updated everything that was pending (i'm a little slack on keeping up with updates). I followed the instructions in this thread, hoping to fox my sound issues, including reinstalling and recompiling galore, but nothing will make Ubuntu see my soundcard again. I also posted to the same forum (link), but no one there has said anything.

In the meantime, thinking maybe now would be a good time to make some backups, I've realized that my CD burner is no longer working. I get a generic error message. Right before any burning starts - "An error occurred while writing". Helpful.

Also, to further confuse things - fusermount no longer wants to work. I use it on a daily basis to mount different machines I have ssh access to via sshfs, and it's worked flawlessly for well over a year. Now, I get this: 'fusermount: mount failed: Operation not permitted'. This happens no matter what I do - it doesn't matter which machine I try to mount, it doesn't work.

I can't help but think this is all related. Bad memory? Flaky harddrive? Linux gurus - what would you do? (Oh, our long time sysadmin quit a couple months ago, so I'm pretty much on my own here...) I'm getting someone else to download and burn me a LiveCD (Gutsy), but that'll take a couple hours. I'm not a linux newbie by any means, but I'm definitely not an expert. I'm a developer; my strengths are not in debugging installs or hardware. That being said, I can check log files (if I know which one to check), google with the best of them and follow directions, but having done all that, I'm now here asking if any linux mefiers have any insights or suggestions? If this was your box - what would you do?

posted by cgg to Technology (18 answers total)

Response by poster: Oh - I forgot to mention: I also have a Windows partition on this box. I tried booting into that - the sound there works fine. I haven't checked the burner there - I guess I should.
posted by cgg at 9:45 AM on April 4, 2008

Is there anything useful in the output of 'dmesg' or in /var/log/messages?
posted by jquinby at 9:47 AM on April 4, 2008

Response by poster: More info:
1) CD burner works fine under windows. Now that I think about it, I honestly can't remember if i've used it under linux. Might be a red herring, I don't know.
2) I've since disabled the soundcard in the BIOS, but that doesn't help any of my other issues.
3) /var/log/messages of note:
nvidia: module license 'NVIDIA' taints kernel.
**** SET: Misaligned resource pointer: f7eee302 Type 07 Len 0
4) dmesg isn't very helpful, because their timestamps are cryptic, but nothing really stands out. Then again, I'm not sure what I'm looking for?
posted by cgg at 10:09 AM on April 4, 2008

Instead of troubleshooting each individual issue (which could take quite a while) it may be easier to back up your important data and re-install once you get that LiveCD. I realize your CD burner is no longer working under Ubuntu but what about using a USB thumb drive or installing an ext3 driver in Windows and using it to read and burn what you need? Have you tried using cdrecord from the command line to see if you can get more meaningful error messages?

I can't tell you what happened. It's a very odd situation you've found yourself in but there should be no reason why running it for a long time (what was it? 4 days?) should cause any of those things. I just checked the uptime on my Linux install and it is at 41 days right now. Did anyone have physical access to the machine while you were gone? Were there any other apps open besides the streaming audio? You're running Ubuntu but did you enable root login graphically and is this the user that you were logged in with? If so, that could cause all sorts of problems including the keyboard and mouse headaches.

Still, you asked what would I do if it was my machine and I'm telling you I would re-install. It shouldn't take longer than 2 hours to re-install and get all your settings back to the way you had them.
posted by pallak7 at 10:21 AM on April 4, 2008

I think I'm with pallak7. It has the smell of a hardware glitch, but you've said that everything works fine under Windows. Assuming your hardware is OK, I'm guessing something - module, set of libraries, who knows - somehow got borked. In the time it takes you to locate and address it, you could have a clean install completed.
posted by jquinby at 11:03 AM on April 4, 2008

The fact that your Windows partition is not experiencing the same problems is puzzling (and annoying :P).

Whenever I have to start chasing multiple 'unrelated' problems... I check the memory. In the past, that's always been the hardest problem to pinpoint, you could potentially save time and help narrow down the problem if you start with MemTest86 first.

Now that I think about it, most Ubuntu LiveCd's already include MemTest86. You should also keep an ear out for your hard drive, you would most likely hear any mechanical problems developing.

That being said... I second the re-install option. I only say this because Ubuntu makes it so easy. Avoid an upgrade, you don't want to carry any issues over into Gutsy.

Unless you run some heavy mods, it's should take less than 2 hours to be up and running again. Which is usually less time than it takes to research and troubleshoot problems.

The fact that your having trouble mounting drives, might be related to your CD burner problem. Does anyone know if data gets lost when fsck kicks in to repair bad sectors?
If so, that might explain some issues. Did you try reinstalling your mount packages?

Keep us informed!
posted by FBXRD at 11:06 AM on April 4, 2008

If you're networked you ought to be able to upgrade without a source CD. Just run:
gksu "update-manager -c -d"
at a terminal window. This will tell Dapper's Update Manager to enable looking for new versions of Ubuntu rather than just upgrading individual components from within the Dapper repositories.
posted by caution live frogs at 11:22 AM on April 4, 2008

what kernel package are you using? After I updated yesterday I ran into a similar issue with some hardware including my PATA hard drive controller not being detected correctly at boot. The box is running Hardy beta, so It might not be a shared issue, though.

I mention this because on the Ubuntu bug tracking site, I found an issue concerning how some of the boot-time details are set up when a new kernel is installed (initramfs, specifically).

I got around it by booting to the previous (working) kernel version using the GRUB menu.
posted by ArgentCorvid at 11:26 AM on April 4, 2008

The surest fix is a reinstall. The last time that installed Ubuntu it took me almost a full day and not two hours. Installing all of the programs that I use that require using apt-get and are not on the livecd, setting all of the program configurations, and tweaking everything so that it is the way I like it takes time. Add in the time for a full backup of your Windows partition and the files in Ubuntu that you want to save and it is going to take you more than two hours to get your computer back the way that you want it.

I would boot from a live Ubuntu cd first and check that everything works. Then copy down the setting for you soundcard and cd that the livecd uses and check those against the settings on your installed version.

If you do decide to install a new version of Ubuntu, you might considering skipping over Gutsy and install the 8.04 beta release of Ubuntu. The full release is supposed to be out sometime this month.
posted by calumet43 at 11:28 AM on April 4, 2008

I would refrain from upgrading to 8.04 (Hardy), especially on your work computer. I'm using it at home and I keep running into all sorts of little issues.

As for the quick re-install, most programs save your preferred settings in the home directory.
Just ctrl-h to see the hidden .files and grab what you want to save. Thus saving you some post-install time. It's better than chasing down all those fonts and scripts for Gimp, or finding all of your Firefox add-ons.
posted by FBXRD at 11:54 AM on April 4, 2008

I forgot to mention that all the programs which you install using Synapatic package manger,apt-get,aptitude have a .Deb file in the following folder

/var/cache/apt/archives

You can use check this folder when you want to see which version or package is installed on your system, and you can make a backup of this folder and install all the packages into another machine.
If your prefer, there is also a GUI program called APTonCD.
posted by FBXRD at 12:01 PM on April 4, 2008

Response by poster: I'm really really confused now. I can live w/o sound. I cannot live without mounting via sshfs. That being said, I've got my LiveCD. I ram memtest for 30 min - it came across 2 errors; I'm not sure how to interpret them exactly, but I figure 2 errors in 30 min of testing isn't bad. Sound works perfectly under the LiveCD, as well. I didn't bother setting up the network connection, that's a pain, and without a sysadmin it's even more painful. I wasn't able to test the burner (b/c I'm booted off the only CD drive), and sshfs/fuse, because well, that's a pain to set up, and w/o a network is useless.

But... here's the really really odd part. I booted from my harddrive as per usual. This time I had my earphones in (no speakers at work). When natilius (or whatever it's called) was kicking in, I heard that little login chime!!! But once everything was up and running, the soundcard is still not there, and playing a wav actually crashed vlc.

I'm dreading a reinstall. Over the last 18? months, i've tweaked my system heavily... I don't want to re-do that. Upgrading Dapper->Edgey->Feisty->Gutsy via apt-get sounds just as painful as starting clean. I'm debating seeing if I can get work to spring for a new harddrive, and doing a clean install on that... if I can't fix this.
posted by cgg at 12:18 PM on April 4, 2008

Sorry, but with MemTest only zero errors will do... but at least you found your problem!
A lot of people recommending letting MemTest complete at least three whole passes before considering the memory "good".

If you found two errors in 30 minutes... that's not good.

Depending on how much RAM you're running with, you should test the memory sticks individually. If you can find the culprit memory stick , remove it and (if you have enough RAM) see how the system runs without the defective RAM.
posted by FBXRD at 12:37 PM on April 4, 2008

That line about the nvidia driver raises red flags for me too. Have you spent some time trying different searches on the ubuntu forums? (especially with terms related to that message?) It's hard to believe that someone hasn't had this problem already.

That said, I'd probably lean towards doing a reinstall too, especially since you're still on Dapper, and the latest stable version is Gutsy. You're missing out on a lot of bug fixes and functionality.

Most of your system tweaks won't be that hard to recover after a reinstall, if you do everything correctly. For starters:

1) backup your home directory (including all those hidden .directories) These contain many of your tweaks, and putting them back in your new home directory will restore lots of your old config.

2) swing through /etc/ and backup some of the config files that you may have modified there as well.

3) you can reinstall all of your apps and libraries by doing the following before reinstalling

$> dpkg --get-selections | grep -v deinstall > ubuntu-files

This will give you a list of all the currently installed programs in the file "ubuntu-files"

After your new install, copy the list over and do the following:
$> sudo apt-get update
$> dpkg –set-selections <> $> sudo dselect

This will go a long way towards making your transition easier. Also, consider setting up seperate partitions for your data (and/or home directories), so that reinstalls and upgrades are easy in the future
posted by chrisamiller at 12:52 PM on April 4, 2008

Response by poster: Ok - memtest is consistently reporting errors at the same locations for the second stick (The 2nd 1GB stick of 2 GB total memory), so that may be the culprit. I've pulled it out, and am using 1GB for the time being. Meanwhile, I've stumbled across an brand new harddrive I'm going to install Gutsy on, as to not trash my current install.

Thanks for all the help!
posted by cgg at 2:53 PM on April 4, 2008

Response by poster: Oh, I should mention that things are still screwy with just the 1 GB stick, but I suspect that's because somewhere, bad memory ended up writing bad data on the hdd? just a guess... Either way, I guess I'm due for an install anyway.
posted by cgg at 2:59 PM on April 4, 2008

I'm not sure how you rebooted - if you tapped the power button to let it shutdown properly, or if you held it down to just switch it off now.

It sounds like the second one due to the fsck at bootup.

I'd install debsums, & ask it to check installed files. That would catch most corrupted files, which might be indicated by the fsck at reboot. Anything that it detects as off, reinstall. (I'm not sure Ubuntu includes md5sums for every single package, unfortunately.)
posted by Pronoiac at 3:30 PM on April 4, 2008

It sounds like you may have updated while the system was on when it went into the weird state, and grub defaulted to the new kernel when you rebooted. To verify, try rebooting, hitting ESC and choosing a kernel back one or two entries, or one that looks familiar.

If that makes things better and you want to try to fix the latest kernel, boot into it and try:

aptitude search linux-restricted-modules-`uname -r`

and verify that package is installed. I could see that fixing your sound, your cd burner, and maybe even the nvidia thing but the fusermount problem may be unrelated.
posted by moift at 4:28 PM on April 4, 2008