All other OS boots, but FC5 = GRUB error 17

[jvroig]

My laptop has been multibooting XP / FC5 / Ubuntu 6.06 / SUSE 10.1 for 2 months already, works perfectly.

This morning, GRUB failed to load at all.

Strange, since I was just using Excel in XP last night and didn't make any system changes at all. Anyway, no problem, I booted from the Fedora CD, linux rescued it, chroot mnt/sysimage, grub-install dev/hda.

Ok, said no errors. device map was also correct, I only had one harddisk anyway.

Reboot. GRUB loads and still has all my settings from before it conked out - including all other OSes, their previous kernels, and the memtest c/o ubuntu, and the other one (same thing) from SUSE. The only thing missing is the custom background image I created. Surprisingly, even the default bg image of FC5 wasn't there. It was just black.

Fine, whatever, I'm surprised my settings where all there at all. I booted FC5, and it gave me GRUB Error 17. Wow.

So I pressed enter, and it returned me back to the OS options in GRUB. I chose Ubuntu, it ran ok. Reboot. Chose SUSE. Ok too. Reboot. Ran XP. Ok too. Reboot, I choose an older Ubuntu Kernel, ran too. I repeated this process for all old kernels I have (I have more than 10 boot options in grub, I like having all old kernels available to me in case a new update screws something up).

Only FC5 (latest kernel, as well as all old kernels) doesn't boot, gives me grub error 17.

I thought maybe there's something wrong with the setting, so I loaded ubuntu, plugged my 1Gb flashdrive, and looked up the backup file that contains the latest grub setting (dated July 17, 2006). It's an exact copy of menu.lst. It's just 10 days ago, and, like the past two months, worked without a flaw.

Anyway, I looked at it and it's exactly what's in GRUB right now. It should work as it has worked for the past two months, except that after the root line, it gives GRUB Error 17.

Anybody has any insight about this?

Thanks,
-JV

[glennzo]

Let's see your grub.conf or menu.lst and output of /sbin/fdisk -l

[jvroig]

Grub.conf won't help you, I already mentioned it's exactly the same as all my backup files of grub, which have been running for two months.

I'm in Windows right now. When I'm finished here, I'll copy the output of fdisk.

If anybody else has any ideas, that would be great. This isn't much of a hassle for me, but that's not the point. The point is something just seemingly went wrong despite the lack of any system changes, and the conventional grub reinstall also didn't work, despite the fact that it's settings are completely identical to the back ups I store.

[jvroig]

Wow, I just rebooted (was using XP), inserted the FC5 cd to try to linux rescue again.

chroot /mnt/sysimage -> grub-install dev/hda -> this produced errors, saying install_device (dev/hda) format is unrecognizeable (say what?!), so no grub will be installed. Okie dokie, grub is there already, anyhoo.

So I rebooted, this time grub didn't even show the OS options, or reflect any of my settings. It just gave me a command prompt.

Geezus. I tried loading windows - had to do it manually step by step in the prompt. God, if I were a noobie to linux, I'd have fainted already. Anyhoo, so I entered (one by one):
rootnoverify (hd0,0) <enter>
chainloader +1 <enter>
boot (hd0,0) <enter>

Ok, XP booted up fine. So my OSes are still there (which really should be the case), so it's really just grub that is crapping out on me.

anyhoo, after chrooting to /mnt/sysimage, here's the output of /sbin/fdisk -l

Disk /dev/hda: 60.0GB, 60011642880 bytes
255 heads, 63 sectors/track, 7296 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device            Boot        Start       End      Blocks        Id          System
/dev/hda1                       1        1912      1535818+        7           HPFS/NTFS
/dev/hda2         *           6045       7296    10056690       83          Linux
/dev/hda3                      1913       6044    33190290       f            W95 Ext'd (LBA)
/dev/hda5         *           4740       5986    10016496       83          Linux
/dev/hda6                      1913       1925      1 04359+     83          Linux
/dev/hda7                      1926       4739    22603423+     8e          Linux LVM
/dev/hda8                      5987       6044       465853+     82          Linux swap / Solaris

Partition table entries are not in disk order

That's it.

I'm already set to delete all linux partitions and reinstall everything. I just dropped by here to see if someone has any insight about this.

Remember, when GRUB crapped out, no system changes were made prior to it. And when it crapped out to just the prompt, no system changes were also made prior to it. No upgrades, no updates, no new installs, nothing. It just quit on me for some reason, and I want to know why before I "fix" (reinstall everything, hehe) it. It's not much of a bother for me, as I use external harddisks to store my critical data and I backup everything quite often (which is why I can comfortably install 4 OS in just a 60 GB laptop drive)

Thanks.

[jvroig]

Just a bump.

A anybody have an idea?

I'm not looking for a solution that would restore GRUB - but if you have one, then that's cool.

What I'm really looking for more is any idea why that would have happened. Before you say "well, a system update could do that..." please, read the initial post, and even the post before this one. I've been multi-booting for two months. Nothing is wrong with grub.conf. And when grub crapped out (in both cases), no system changes were made at all. No updates. No installs. I was just using XP for MS Excel (in both cases). Well, Firefox did update itself without my consent in XP (well, downloaded the update then just informed me it has finished, I was like, "WTF?!", then asked me if I want to install the update now, and I was like "WTF?!! you wasted my bandwidth for that crap, you better believe I'm gonna install you"). And in the first case, a new version of ZoneAlarm (also in XP, obviously) was installed. But after the install, I rebooted properly. Continued working on Excel (XP still), then shut down my laptop and went to sleep. The morning after that was when GRUB suddenly didn't want to load, so I had to linux resuce through FC5 CD... and that's how everything started. Pretty normal, nothing going, but things just got out of whack - at least GRUB and FC.

Now, even though GRUB only shows a prompt, like I said, as long as I manually enter parameters line by line, the OSes still load - except of course for FC5, still the same GRUB error 17 after the root line. (Again, don't ask for grub.conf. I already checked, it's still exactly the same as it was 10 days ago in my backup grub.conf, which also is essentially the same as for the past two months - except for the kernel versions, of course - so there is nothing in it.

Pretty cool turn of events, eh?

[jvroig]

Dudes, my problem happened again. I just reinstalled a minimal Fedora installation, played around with it for a while. Then I worked on Windows XP again.

And guess what, when I restarted my laptop, GRUB was messed up again. Just gave me a command prompt, and again just like before I could still start Windows XP manually through

# rootnoverify (hd0,0)
# chainloader +1
# boot

I thought, gedemmit, is XP messing up GRUB, or my other OS partition?!
But if that were the case, then lots of peeps here would have related with me, since a lot dual boot with XP. So I don't think it's that.

I'm running tests right now, to see exactly what I do with XP when GRUB gets screwed.

Oh yeah, when I tried grub-install again through linux resuce, it failed, saying unsupported format. I'm pretty sure that means Fedora is unbootable again like before.

I do no updates in Windows (at least, none that ask for my permission, anyway, and I have automatic updates turned off, and I configured ZoneAlarm to not allow it). I also didn't tinker with any system settings.

Partition Magic 8.0 is installed, and it seems a bit screwy. I'm thinking this might be the culprit.

I'll report back here when I get this sorted out.

If any of you guys have any ideas, just holler.

Regards,
-JV

[jvroig]

Bump.

So far, it's been a week. I installed a minimal FC again, this time used a primary partition instead of a logical one.

Still doing all my work in XP since I didn't want to install and configure all my linux OS, but so far the error didn't happen again. Grub is still ok.

I doubt it was a logical partition problem though, never heard of anything like that before.

Any ideas? I don't want to spend half a day installing FC5, Ubuntu 6.06 and SUSE 10.1, then half a day more configuring them all and installing servers and copying my files into them if for some reason the error happens again.

Thanks in advance,
-JV

[bob]

No real answers for you, but how many primary partitions did you create? Max is 4, I believe. In any case, have you gone here: http://www.gnu.org/software/grub/ and file a bug on the problem? You know that all your distros are intact and no changes to them could have caused it. You questioned whether PartitonMagic might be involved, but if it's not that, then Grub itself is borking and the developers need to know when things go bad.

[d3viant]

The only thing I can think of is that somehow your partition order has messed up.

The grub background image is loaded from the /boot folder on your FC5 partition, so if it loses this image, then it isnt finding that partition at all. Look in your /boot/grub/grub.conf file and try playing round with the root (hdx,x) numbers. Your partition number (the 2nd one) should be one number less than the output from fdisk -l (because grub starts from 0, not 1)

[jvroig]

@Bob

There's no problem with the primary partitions. Yep, 4 is max. When the errors happened, I only used two primary partitions, and the logical partitions contain Ubuntu and FC5.

I also used logical when the error happened again. Now that I switched to creating 4 primary partitions (since I have only 4 OS anyway), the problem doesn't recur (but it's just one week).

I don't think it is a logical partitioning problem since I've been working with the same setup for 2 months (but hardly touching XP, except for the times I used XP to watch one whole season of Alias in DVD). That it just "broke" out of the blue is insane, and if I didn't experience it personally and just read about it, I would probably insist that "the user did something wrong, obviously." Alas, I'm no noob at this, I think I've installed, reinstalled, and configured several Linux distros in different machines more than I can count on my fingers and toes. I'm just so stumped how something that worked flawlessly for 2 months can just die on me in the absence of changes, updates, installs, nada!

I woul have filed a bug report already like all decent users. Unfortunately, now that I failed to replicate the error (for what would have been the 3rd time in 2 weeks had it happened again), I don't know what to tell them, since they'll ask me (and rightly so) to see if I can duplicate it consistently. As of now, I can't. I'd love to reformat FC again (since it doesn't have any files at all, I work in XP these days until I have this all sorted out) and use logical partitions again and see if that will get messed up again, but I have some deadlines to meet, and if I don't meet them, the client will suffer some monetary penalties from the government - internal revenue really take their job seriously I guess.

[bob]

Yeah, it's another one of those things that we may never really know what happened or why. If this was XP, we'd just blame the instability of the OS, but linux.... ?

[jvroig]

Originally Posted by d3viant

The only thing I can think of is that somehow your partition order has messed up.

The grub background image is loaded from the /boot folder on your FC5 partition, so if it loses this image, then it isnt finding that partition at all. Look in your /boot/grub/grub.conf file and try playing round with the root (hdx,x) numbers. Your partition number (the 2nd one) should be one number less than the output from fdisk -l (because grub starts from 0, not 1)

Yeah, that hit me too. Since the problem was the partition couldn't be mounted (only the FC partition), and then eventually grub (also in that partition) conked out... I was thinking "damn, something caused my logical partition (just the FC one) to really mess up". What do you think? Everything else was intact, just that particular logical partition, and it happened again in just a few days right after I formatted everything except the XP partition and used another LOGICAL partition (as before) to test out the error and it happened like clockwork after using XP. Now, using a PRIMARY partition, and using XP almost all the time, grub and the FC partition work fine. (But let's not lose sight of the fact that, even with the logical partitions, it worked fine for 2 MONTHS, until out of the blue it just decided to die).

Can't be XP sabotaging my laptop right? After all, a ton of folks dual boot with XP, and I'm sure more than just a handful also use logical partitions, so the fact that it seems none of the thousands of users can relate means it's not about XP this time.

But since I'm sure the grub.conf entries are correct (aside from the fact that every other OS boots properly using the same configuration, I also checked my backup grub.conf, I always keep one, came in handy when this happened), then could it be that only the FC5 logical partition changed? I've been using computers for just a decade, and Linux much shorter than that, but I've never heard that happen before.

Weird huh? I almost wish I will find out that it is part of Microsoft's hidden "destroy the competition" embedded program in XP, but c'mon, let's give MS a break, it's most probably not that at all.

[jvroig]

Originally Posted by bob

Yeah, it's another one of those things that we may never really know what happened or why. If this was XP, we'd just blame the instability of the OS, but linux.... ?

Well, doesn't seem like the fault of Linux at all (wow, not the fault of XP, and not the fault of Linux... man, who to blame, huh? ).

They all worked fine. And there were no updates, system changes, install, etc. And it worked fine for 2 whole months of very active use. So I find it highly unlikely to be an OS fault, whether XP, FC, Ubuntu, or SUSE.

But yeah, I guess another one of those things that we may never really know what happened. Bummer.

What I'm not much of an expert at is in hardware. Can any hardware guru around here tell me if this could have been caused by hardware issues (maybe my HD wants a replacement? My laptop's barely two years old, and this is a good brand, an HP Pavilion).

Thanks guys.

[bob]

Never a bad idea to go to the manufacturer's site and get the tools to test a drive, since most of them have a 3-yr warranty and you can save your data if you have some idea that things are going bad.

One other thing - let's say you had Ubuntu on first with it's LVM and VolGroup00, then comes Fedora and the LVM for that is VolGroup01. Now, due to a glitch in a kernel update for Fedora or grub or whatever and the VolGroup is reset to the default of 00. Grub has now lost the map and since the meat of grub (from the last distro installed) is on Fedora, it's not functioning at all. Since you mentioned that grub had completely borked, I think that's a possibility.

[jvroig]

Originally Posted by bob

Never a bad idea to go to the manufacturer's site and get the tools to test a drive, since most of them have a 3-yr warranty and you can save your data if you have some idea that things are going bad.

One other thing - let's say you had Ubuntu on first with it's LVM and VolGroup00, then comes Fedora and the LVM for that is VolGroup01. Now, due to a glitch in a kernel update for Fedora or grub or whatever and the VolGroup is reset to the default of 00. Grub has now lost the map and since the meat of grub (from the last distro installed) is on Fedora, it's not functioning at all. Since you mentioned that grub had completely borked, I think that's a possibility.

Yup, but FC came first (VolGroup00).

And the error happened at least a few days after an update. This is my main work laptop, so it's used at least 8 hours of every day.

I just did a linux-rescue grub-install after all OSes were installed to use the FC grub instead of SUSE (which I installed last, right after Ubuntu). That was 2 months ago.

Anyhoo, I just got a second laptop. I'll remove XP in my main laptop, and then I'll try to replicate the error in the second laptop. I'll file a bug report when I can consistently make the error happen.

Thanks guys.