Failed dist upgrade now worry over lvm data

Fettrix · 15th January 2008, 10:09 PM

hi all,

I had a debian 3.1 install that I tried to upgrade to 4.0. Now as it turns out the OS HDD (/dev/hda) has something seriously wrong with it so I need to replace the drive.

My concern is for the data I have on my LVM set up which is as follows:

/dev/hdc1 - LVM
/dev/hdd1 - LVM
/dev/hde1 - LVM
/dev/hdf1 - LVM
/dev/hdg1 - LVM

Would it be safe to replace hda with a new HDD, and do a fresh install on that? If so how would I be able to keep the LVM data as it currently is?

Any help would be great as my last back up was 27/9/7 an I have since added more things which i dont have a copy of.

Many thanks for any info

Fett.

cleary · 15th January 2008, 10:40 PM

Quote:

Originally Posted by Fettrix

...
My concern is for the data I have on my LVM set up which is as follows:

/dev/hdc1 - LVM
/dev/hdd1 - LVM
/dev/hde1 - LVM
/dev/hdf1 - LVM
/dev/hdg1 - LVM

...

this is not enough info - lvm consists of physical volumes, volume groups and logical volumes.
LVM stands for Logical Volume Management - that refers to the process of managing the the three items mentioned above.

Can you paste the output of

Code:

lvdisplay

and

Code:

vgdisplay

Fettrix · 15th January 2008, 11:02 PM

Quote:

Originally Posted by cleary

this is not enough info - lvm consists of physical volumes, volume groups and logical volumes.
LVM stands for Logical Volume Management - that refers to the process of managing the the three items mentioned above.

Can you paste the output of

Code:

lvdisplay

and

Code:

vgdisplay

I wish i could paste the info, but as it is, the machine no longer connects to my network due to the corrupted primary hdd. I have tried /etc/init.d/ssh restart but get command not found, which i get when i try to run most things on the box.

I have taken some photos though. Sorry for the blurr.

Code:

lvdisplay

Click to view full-sized image!
Hosted by UGBox Image Store

Code:

vgdisplay

Click to view full-sized image!
Hosted by UGBox Image Store

DaveQB · 16th January 2008, 10:24 AM

Physical Volumes in an LVM has metadata on them. So simply re-install the OS onto a new disk at /dev/hda and then bring your LVM back up.

done

Fettrix · 16th January 2008, 11:12 AM

Quote:

Originally Posted by DaveQB

Physical Volumes in an LVM has metadata on them. So simply re-install the OS onto a new disk at /dev/hda and then bring your LVM back up.

done

OK but going via those pics, (don't forget I am a total noob at this) it seems that there are 6 metadata area's, does this translate into 6 hdd's? If so then wont re installing the OS on a new drive stuff up my LVM as it is?

If i am correct and I di have the LVM spread across the 6 HDD's, how can i remove the dying HDD from it?

MWP · 16th January 2008, 12:40 PM

Whats does your /etc/fstab contain?

Surely you are not just running one big partition across all drives

Fettrix · 16th January 2008, 1:03 PM

Quote:

Originally Posted by MWP

Whats does your /etc/fstab contain?

Surely you are not just running one big partition across all drives

Will see what that fstab thingy has in it when i get home tonight.

As for the partition I am on hdc - hdg as that was only being used by windows for file storage, and at the time i set it up, was the simplest method.

DaveQB · 16th January 2008, 2:57 PM

Quote:

Originally Posted by Fettrix

OK but going via those pics, (don't forget I am a total noob at this) it seems that there are 6 metadata area's, does this translate into 6 hdd's? If so then wont re installing the OS on a new drive stuff up my LVM as it is?

If i am correct and I di have the LVM spread across the 6 HDD's, how can i remove the dying HDD from it?

Only mess things up if the OS is in the LV. If the OS is on its own drive, (hda?) and no part of any LV, then it has no baring on the LV and your sweet.

As for the latter question.... you cant, its not a RAID where you can hot remove and hot add, its LVM. Best to run LVM ontop of a RAID.

DaveQB · 16th January 2008, 2:58 PM

Quote:

Originally Posted by MWP

Whats does your /etc/fstab contain?

Surely you are not just running one big partition across all drives

Yeah it is!!

Talking to Fetta offline and it seems hda has the OS and hdc-hdg have the LV. So he is in the clear as hda appears to be faulty.

MWP · 16th January 2008, 6:36 PM

Ouch... good luck with that

-=N0N@ME420=- · 16th January 2008, 6:44 PM

Quote:

Originally Posted by DaveQB

Yeah it is!!

Where's the redundancy?

f3n1x · 16th January 2008, 11:23 PM

Sorry to drag the thread semi offtopic but what the hell is with the massive almost magnetic attraction to raid, everything has to be redundant, i find this to be something of a wank factor in non-mission critical environments.

I understand people using raid to provide high-availability and such, i also understand it from the disk io/performance position, what i don't 'get' is this almost fabled faith in raid arrays for prevention of data loss.

I've come across a number of it "professionals" who have discussed how they have a RAID array and can therefore get away with only a monthly backup, my mind boggles at the mere concept.

A well thought out and properly implemented backup plan in my experience is orders of magnitude more important that disk redundancy, and in my working life i've seen a significant number of jobs, contracts and clients saved and lost on good, bad and ugly backup implementations.

As much as this may be an unpopular opinion, i feel it to be a valid one.

MWP · 17th January 2008, 3:11 AM

Depends...

RAID0 and RAID1 are useful in a home environment.
Other types of RAID and LVM are overkill though and normally just complicate setups too much (like the above problems).

I keep my docs, work stuff, photos on a RAID1 drive.
Everything else is on plain XFS partitions (the linux server has 3x120Gb and 3x250Gb drives).
If my OS install falls over for whatever reason, its very easy to get everything going again.

MrSnuffy · 17th January 2008, 12:16 PM

eh?! LVM is very useful in any environment.

Not everyone needs or wants redundancy for drive failures or the extra speed offered by RAID0.

I just have a bunch of disks in a volume group.

DaveQB · 17th January 2008, 1:07 PM

Quote:

Originally Posted by -=N0N@ME420=-

Where's the redundancy?

Yeah exactly. A bit concerning. 1 disk goes down and the whole thing is lost!!!

15th January 2008, 10:09 PM	#1
Fettrix Member Join Date: Jul 2002 Location: Eastern, VIC, (Knox) Posts: 1,357	Failed dist upgrade now worry over lvm data hi all, I had a debian 3.1 install that I tried to upgrade to 4.0. Now as it turns out the OS HDD (/dev/hda) has something seriously wrong with it so I need to replace the drive. My concern is for the data I have on my LVM set up which is as follows: /dev/hdc1 - LVM /dev/hdd1 - LVM /dev/hde1 - LVM /dev/hdf1 - LVM /dev/hdg1 - LVM Would it be safe to replace hda with a new HDD, and do a fresh install on that? If so how would I be able to keep the LVM data as it currently is? Any help would be great as my last back up was 27/9/7 an I have since added more things which i dont have a copy of. Many thanks for any info Fett. __________________ * Google * Trades * DVD’s * Ç¥BËRÇÔRË * Last.fm * Steam Games * Indie Count * __________________ "Two there should be; no more, no less. One to embody power, the other to crave it."

16th January 2008, 10:24 AM	#4
DaveQB Member Join Date: Apr 2002 Location: Narellan, Sydney Posts: 2,659	Physical Volumes in an LVM has metadata on them. So simply re-install the OS onto a new disk at /dev/hda and then bring your LVM back up. done __________________ AMD Phenom X6 1090t @3,600 Mhz (stock) \| GA 870A-UD3 \| Corsair 16GB DDR3 1333 \| GTS 250 stock \| 6TB of storage \| Running Linux Mint 13 \| trades \|\| My Blog \|\| Linux User 417163 \|\| Twitter \|\| Vegan Australia

16th January 2008, 12:40 PM	#6
MWP Member Join Date: Jun 2001 Location: Adelaide Posts: 4,636	Whats does your /etc/fstab contain? Surely you are not just running one big partition across all drives __________________ MWP PC Database AdminCrystalfontz CrystalControl2 Author - Offical Forums Motoring Worklog - Toyota V8 RA28 Celica SA Classic Celica Club Member - Website

16th January 2008, 6:36 PM	#10
MWP Member Join Date: Jun 2001 Location: Adelaide Posts: 4,636	Ouch... good luck with that __________________ MWP PC Database AdminCrystalfontz CrystalControl2 Author - Offical Forums Motoring Worklog - Toyota V8 RA28 Celica SA Classic Celica Club Member - Website

16th January 2008, 11:23 PM	#12
f3n1x Member Join Date: Mar 2003 Location: Armadale, Melbourne Posts: 1,653	Sorry to drag the thread semi offtopic but what the hell is with the massive almost magnetic attraction to raid, everything has to be redundant, i find this to be something of a wank factor in non-mission critical environments. I understand people using raid to provide high-availability and such, i also understand it from the disk io/performance position, what i don't 'get' is this almost fabled faith in raid arrays for prevention of data loss. I've come across a number of it "professionals" who have discussed how they have a RAID array and can therefore get away with only a monthly backup, my mind boggles at the mere concept. A well thought out and properly implemented backup plan in my experience is orders of magnitude more important that disk redundancy, and in my working life i've seen a significant number of jobs, contracts and clients saved and lost on good, bad and ugly backup implementations. As much as this may be an unpopular opinion, i feel it to be a valid one. __________________ f3n.org\|systems admin, graphics & foss software Canon Eos 40D. Canon 50mm f1.8. Tamron SP AF28-75mm F/2.8 XR Di LD Aspherical (IF). Sigma EF 500 DG Super ETTL Flash(Broken! :/)

17th January 2008, 3:11 AM	#13
MWP Member Join Date: Jun 2001 Location: Adelaide Posts: 4,636	Depends... RAID0 and RAID1 are useful in a home environment. Other types of RAID and LVM are overkill though and normally just complicate setups too much (like the above problems). I keep my docs, work stuff, photos on a RAID1 drive. Everything else is on plain XFS partitions (the linux server has 3x120Gb and 3x250Gb drives). If my OS install falls over for whatever reason, its very easy to get everything going again. __________________ MWP PC Database AdminCrystalfontz CrystalControl2 Author - Offical Forums Motoring Worklog - Toyota V8 RA28 Celica SA Classic Celica Club Member - Website Last edited by MWP; 17th January 2008 at 3:15 AM.

17th January 2008, 12:16 PM	#14
MrSnuffy Member Join Date: Jun 2001 Location: Hobart Posts: 1,235	eh?! LVM is very useful in any environment. Not everyone needs or wants redundancy for drive failures or the extra speed offered by RAID0. I just have a bunch of disks in a volume group.