where else are drives called other then fstab?

[Red Squirrel]

I got a failed HDD but most of the real stuff is on the non failed ones, so I just removed it completely and I went to remove it from fstab but noticed its not even in there, but yet something is calling it as it refuses to boot up because it can't find it.

Where else could I look? I don't even remember putting it somewhere else then fstab since I installed it maybe like a month ago, and went through the same ctrl+d screen nightmare. They really gotta change that so it lets you boot anyway. Sure some services may fail but if the core OS can at least be started it would make it semi usable depending on situation. (in this case its not so bad as the drive was not a high production drive other then my UO shard dev VM)

 

[kamper]

Don't bother saying what os you're using. We like guessing.

 

[Red Squirrel]

FC5

You'd think they'd make these thinks consistent from distro to distro.

 

[Nothinman]

fstab is all I can think of right now, but if the disk was part of a LVM volume group I don't know what it going missing would cause.

 

[Red Squirrel]

Was just a regular disk. I just don't get how it was even getting mounted if its not part of fstab. It was being mounted to /data3 and theres no entry for /data3 in fstab.

 

[Brazen]

The only other possibility is it was mounted in a custom boot up script. I believe startup scripts go in the /etc/rc#.d directories, depending on when in the startup process you want the script to run. You could try running this to see if you get any response:

grep -r "data3" /etc/rc*

 

[Red Squirrel]

Hmm even if it was called externally it would fail to boot? what is the best way then to auto mount drives without it failing the whole boot process if it can't mount? Though I don't recall making a script for that but I'll have to double check when I go home.

 

[Crusty]

Just because YOU didn't make the script doesn't mean that someone else didn't, like FC5. Especially if you used some custom gui tool to do any kind of system configuration.

 

[Brazen]

Originally posted by: RedSquirrel
Hmm even if it was called externally it would fail to boot? what is the best way then to auto mount drives without it failing the whole boot process if it can't mount? Though I don't recall making a script for that but I'll have to double check when I go home.

I haven't really had much experience with failed drives to know what will or will not stop a boot, but I would think no matter where you put it, even in fstab, as long as system files were not on it, then the computer would still boot. If nothing else, even if you don't find anything bry grepping in /etc/rc* you could do the grep on /etc and see if anything comes up. It will take a long, looooong time to complete though.

 

[Red Squirrel]

Linux is super fussy about that. If it can't find one of the drives in FSTab it just crashes at the start with that dreaded ctrl+d screen.

Anyway I managed to get it going. I feel like retard. There was a space in the FStab then the drive was there, but guess my eyes totally wandered off it and I never noticed it last night when I looked. I was still in a state of panic then. I went back in fstab and sure enough it was there. The dark blue did not help either. So I commented it out for now and system booted up. Mysql failed since the databases were on that drive, but I'll worry about that once I get the new drive. Have no clue how I'll go about restoring a .sql to a broken mysql data structure though, but I'll post if I have issues then.

 

[Nothinman]

Linux is super fussy about that. If it can't find one of the drives in FSTab it just crashes at the start with that dreaded ctrl+d screen.

Uh, no. Linux isn't fussy at all, that's the init scripts for your distribution.

The dark blue did not help either.

:set background=dark

 

[techbrain55]

First off all you did not mentioned what OS you are using. All you have to do is edit your fstab. For my windows drives I created a folder in my home called drives and folders called winxp and dataHd. Then in fstab i point to those new folders which I still can see in nautilus but they are not on the desktop.
Regards,
Techbrain55

 

[Brazen]

Originally posted by: techbrain55
First off all you did not mentioned what OS you are using. All you have to do is edit your fstab. For my windows drives I created a folder in my home called drives and folders called winxp and dataHd. Then in fstab i point to those new folders which I still can see in nautilus but they are not on the desktop.
Regards,
Techbrain55

He's using Fedora Core 5.

 

[JohnOfSheffield]

Originally posted by: RedSquirrel
I got a failed HDD but most of the real stuff is on the non failed ones, so I just removed it completely and I went to remove it from fstab but noticed its not even in there, but yet something is calling it as it refuses to boot up because it can't find it.

Where else could I look? I don't even remember putting it somewhere else then fstab since I installed it maybe like a month ago, and went through the same ctrl+d screen nightmare. They really gotta change that so it lets you boot anyway. Sure some services may fail but if the core OS can at least be started it would make it semi usable depending on situation. (in this case its not so bad as the drive was not a high production drive other then my UO shard dev VM)

Some distros have a strange way of determining what is what, i know Ubuntu does, it goes by drive number/partition numbers and it can be pure hell to even get it to boot since you can't determine the partition order without it booting and so you can't fix your fstab offline either, i don't run FC5 so i don't know much about how it does that, if it's the same way that Ubuntu does then that may be problematic, if it's the way SuSE does it (a combo with a dynamic fstab) then it might be easier, or you could just edit the fstab by hand by booting from a live cd and checking out your partitions and assigning them the the correct mount points in the old fashioned way.

Could you do a cat /etc/mtab and post it? With the failed drive in there of course, as it won't boot without it.

 

[Nothinman]

Some distros have a strange way of determining what is what, i know Ubuntu does, it goes by drive number/partition numbers and it can be pure hell to even get it to boot since you can't determine the partition order without it booting and so you can't fix your fstab offline either,

They all use the same drivers so the device names will be the same. The order of the devices can change though with module loading order and other timing issues which is why Ubuntu now uses UUIDs to identify filesystems. And with something like that it's simple to boot a LiveCD and fix any problems.

 

[JohnOfSheffield]

Originally posted by: Nothinman

Some distros have a strange way of determining what is what, i know Ubuntu does, it goes by drive number/partition numbers and it can be pure hell to even get it to boot since you can't determine the partition order without it booting and so you can't fix your fstab offline either,

They all use the same drivers so the device names will be the same. The order of the devices can change though with module loading order and other timing issues which is why Ubuntu now uses UUIDs to identify filesystems. And with something like that it's simple to boot a LiveCD and fix any problems.

Well, you may call it simple, i call it crap, a dynamic fstab is a ten times better fix than having to find a 15 digit number out of several and try to figure out what partition that is, then you do, reboot and the god damn partition fifteen digit number has changed.

Look at how SuSE does it and tell me that that isn't easier, or how it used to be done with a static fstab with your sda1, sdb2 and so on. I think they ruined Ubuntu in their chase for user friendliness.

It's just my opinion though and i'm probably a bit oldschool on this matter.

 

[Nothinman]

Well, you may call it simple, i call it crap, a dynamic fstab is a ten times better fix than having to find a 15 digit number out of several and try to figure out what partition that is, then you do, reboot and the god damn partition fifteen digit number has changed.

The point of a UUID doesn't change unless you recreate the filesystem, if your UUIDs change from a reboot you've got other issues. And I don't want anything editing fstab for me, there's no way some shell script is going to do the right thing every time.

Look at how SuSE does it and tell me that that isn't easier, or how it used to be done with a static fstab with your sda1, sdb2 and so on. I think they ruined Ubuntu in their chase for user friendliness.

I haven't looked at SuSE in a while since every time I tried in the past I got frustrated with all of their "friendly" crap doing stuff behind my back without a good way to see what was going on.

Frankly I'd just recommend that most people use LVM these days, it's a lot more flexible and you don't have to worry about what order your block devices come up in as long as they're all up by the time the lvm scripts run in your initramfs.

 

[Red Squirrel]

how does LVM work? I've seen fstabs where the partitions are actually refered to as labels. how do you set that up, and is there a way to do it with ext3 without having to reformat? May be worth looking at, sounds nice.

Also is there a way to get a dump of all the superblocks on all drives? Aparantly to repair a drive with FS damage you need to know a backup super block, so think I'll keep a record of all my drives/partitions.

 

[Nothinman]

how does LVM work?

Essentially you set a partition as a LVM physical volume, add it to a volume group and then create a volume from the pooled space in that volume group. You then get a block device named /dev/mapper/<vg-name>-<vol-name> and a symlink in /dev/<vg-name>/<vol-name> so you can use either naming convention to refer to it.

You can change the label on an ext2/ext3 filesystem at any time with e2label. Although you might have to do some fiddling with udev or just reboot to get it to create the symlink in /dev/disks/by-label.

Also is there a way to get a dump of all the superblocks on all drives? Aparantly to repair a drive with FS damage you need to know a backup super block, so think I'll keep a record of all my drives/partitions.

From the mke2fs man page:

-n causes mke2fs to not actually create a filesystem, but display what it would do if it were to create a filesystem. This can be used to determine the location of the backup superblocks for a particular filesystem, so long as the mke2fs parameters that were passed when the filesystem was originally created are used again. (With the -n option added, of course!)

 

[DaiShan]

Originally posted by: RedSquirrel
Linux is super fussy about that. If it can't find one of the drives in FSTab it just crashes at the start with that dreaded ctrl+d screen.

Anyway I managed to get it going. I feel like retard. There was a space in the FStab then the drive was there, but guess my eyes totally wandered off it and I never noticed it last night when I looked. I was still in a state of panic then. I went back in fstab and sure enough it was there. The dark blue did not help either. So I commented it out for now and system booted up. Mysql failed since the databases were on that drive, but I'll worry about that once I get the new drive. Have no clue how I'll go about restoring a .sql to a broken mysql data structure though, but I'll post if I have issues then.

No, not the case, if the drive is necessary (root partition, part of an LVM group etc) then it will certainly fail to boot, just as it would in Windows, or Solaris, or HP-UX. FC and RHEL main have used LVM as the default partition type for quite some time. Are you absolutely positive this is not part of an LVG? What does your /etc/fstab file say?

 

[JohnOfSheffield]

Originally posted by: Nothinman

Well, you may call it simple, i call it crap, a dynamic fstab is a ten times better fix than having to find a 15 digit number out of several and try to figure out what partition that is, then you do, reboot and the god damn partition fifteen digit number has changed.

The point of a UUID doesn't change unless you recreate the filesystem, if your UUIDs change from a reboot you've got other issues. And I don't want anything editing fstab for me, there's no way some shell script is going to do the right thing every time.

Look at how SuSE does it and tell me that that isn't easier, or how it used to be done with a static fstab with your sda1, sdb2 and so on. I think they ruined Ubuntu in their chase for user friendliness.

I haven't looked at SuSE in a while since every time I tried in the past I got frustrated with all of their "friendly" crap doing stuff behind my back without a good way to see what was going on.

Frankly I'd just recommend that most people use LVM these days, it's a lot more flexible and you don't have to worry about what order your block devices come up in as long as they're all up by the time the lvm scripts run in your initramfs.

I'm not talking about a shell script when i'm mentioning a dynamic fstab, i'm talking about hotpluggin it in via the devd and assigning it based on devid and partid, tell me how that will fail and tell me why it never does in SuSE, i can hotplug a drive and the fstab will assign a new devid and partid unless it contains the OS, of course.

oh, i should have read your entire post before i wrote that i guess, most unix people like me just partition their drives, an LVM would suffer even more from a failed drive though, unless it's in a non-redundant array, but most people don't have that today, it seems to me that LVM is prett much just drive spanning without the benefit of read/write speed (that doesn't really matter in home computers anyway but that is OT).

But i agree with you, with backups of your stuff (remember folks, this is needed even if you are on a redundan array).

I ls'd a folder yesterday, did an rm -rf * and forgot that i wasn't in the folder i just listed, wiped my drive, it took me hours to fix it, testdisk to the rescue, damn fine piece of software it is.

I'd had to spend hours resotoring the latest backup if i had not been aware of Testdisk.

 

[JohnOfSheffield]

Originally posted by: Nothinman

how does LVM work?

Essentially you set a partition as a LVM physical volume, add it to a volume group and then create a volume from the pooled space in that volume group. You then get a block device named /dev/mapper/<vg-name>-<vol-name> and a symlink in /dev/<vg-name>/<vol-name> so you can use either naming convention to refer to it.

You can change the label on an ext2/ext3 filesystem at any time with e2label. Although you might have to do some fiddling with udev or just reboot to get it to create the symlink in /dev/disks/by-label.

Also is there a way to get a dump of all the superblocks on all drives? Aparantly to repair a drive with FS damage you need to know a backup super block, so think I'll keep a record of all my drives/partitions.

From the mke2fs man page:

-n causes mke2fs to not actually create a filesystem, but display what it would do if it were to create a filesystem. This can be used to determine the location of the backup superblocks for a particular filesystem, so long as the mke2fs parameters that were passed when the filesystem was originally created are used again. (With the -n option added, of course!)

Why is it possible to have disks-by-label and not partitions-by-label in an Ubuntu fstab, i know you are probably not the man to actually ask about this since you're a Debian man and it's not done the same way.

Maybe i should just go debian and bug you about everything... just kidding.

 

[Nothinman]

I'm not talking about a shell script when i'm mentioning a dynamic fstab, i'm talking about hotpluggin it in via the devd and assigning it based on devid and partid, tell me how that will fail and tell me why it never does in SuSE, i can hotplug a drive and the fstab will assign a new devid and partid unless it contains the OS, of course.

That works on any distro with udev, look at all of the symlinks under /dev/disk.

oh, i should have read your entire post before i wrote that i guess, most unix people like me just partition their drives, an LVM would suffer even more from a failed drive though, unless it's in a non-redundant array, but most people don't have that today, it seems to me that LVM is prett much just drive spanning without the benefit of read/write speed (that doesn't really matter in home computers anyway but that is OT).

Most is probably overstating things, for example Fedora has used LVM by default for a long time now so just about every Fedora user uses LVM instead of normal partitions. I really wish Ubuntu would've done that too because then they could've avoid the whole conversion to UUID mess.

Why is it possible to have disks-by-label and not partitions-by-label in an Ubuntu fstab, i know you are probably not the man to actually ask about this since you're a Debian man and it's not done the same way.

Actually it is done the same way, the way device nodes are created on every distro is handled by udev. The names of some of the created symlinks like /dev/cdrom may be different but that's about it. And you can't do disks-by-label as there's no way to label an entire disk, labels only apply to individual filesystems.