Switching one of my disks to hdc from hdb
Hi,
since some of you told me my Setup with two stripped disks on hda and hdb was broken, I would like to put my second disk (/dev/hdb) on the second Controler (/dev/hdc) instead of my DVD Drive. Now my question is: What do I have to change, in order not to distroy my linux installation in the process ?? I can realize fstab will have to be changed but else ? Anything in some lvm files ? Surely ! Please help me
Thank you
pwk.linuxfan
Quoted 4 lines · Peter Kirk
since some of you told me my Setup with two stripped disks on hda and hdb was broken, I would like to put my second disk (
/dev/hdb) on the second Controler (/dev/hdc) instead of my DVD Drive. Now my question is: What do I have to change, in order not to distroy my linux installation in the process ??
Well, as I don't know your current configuration to well, it's just pocking in the dark ...
If your drives are PVs in one VG, that's not the root system, all you need to do is:
- do a clean shutdown
- change drives
- boot again
- run vgscan (should find your PVs and pack them into one VG) (*)
- run vgchange -a y (should activate your VG) (*)
- mount all LVs in the VG
For me the two steps marked with (*) are done in a initrd and the system would even support changing the root's LV, as long as the /boot stays untouched.
My setup is like this:
/dev/hda1: 64MB /boot/dev/hda2: rest MD0 disk1 (raid 1)
/dev/hdc1: 64MB unused/dev/hdc2: rest MD0 disk2 (raid 1)
/dev/hde1: PV1/dev/hdg1: PV2
/dev/hdi1: PV3/dev/hdk1: PV4
/dev/md0: PV0
VG0: PV0
VG1: PV1 PV2
VG2: PV3 PV4
hda and hdc are 40GB IBM drives, hd[e-h] were 60GB Maxtor drives.
I'm running a full system out of different LV in VG0 (even root is a LV, only /boot is extra, with inird)
Lately I get a 3ware Escalade controller and moved hd[e-h] there. All I had to do was:
- change initrd to load the 3ware driver (run lilo to install new initrd)
- shutdown, install card, move drives, (remove additional IDE controllers), reboot
- initrd detected all PVs and constructed all my VGs, system still ran fine (you might to have to copy the newly detected configuration from /initrd/etc/lvm to /etc/lvm after boot)
(Actually I did it a little different: shutdown, install 6 new 80GB drives, make a RAID5 with this drives, shutdown, add two drives out of VG1, boot (VGs detected on sd[gh]), make PV5 out of RAID5 array, make VG3 with just this PV5, copy all the data from VG1's LV to LVs on VG3, shutdown, remove drives, add drives from VG2, do the same as with VG1, shutdown, remove drives, boot, migration done, system works, all fine.)
Goetz.
Am Mittwoch, 6. Juni 2001 23:26 schrieben Sie:
Quoted 9 lines · Peter Kirk
Quoted 5 lines
since some of you told me my Setup with two stripped disks on hda and hdb was broken, I would like to put my second disk (
/dev/hdb) on the second Controler (/dev/hdc) instead of my DVD Drive. Now my question is: What do I have to change, in order not to distroy my linux installation in the process ??Well, as I don't know your current configuration to well, it's just pocking in the dark ...
Here it comes:
I have two disks, currently on hda and hdb.
hda:
10MB of /boot
1GB for SWAP
the lot (18GB) in LV with hdb
hdb:
19GB in LV with hda
The LV with 18GB+18GB has
{
/
/usr
/home
all of these are stripped, so they are like raid0 (or should be) /home/stuff is not stripped, because it is on hdb where hda has boot and swap
}
pwk@notch:~/transcendent/mki/unstable > cat /proc/lvm/global
LVM driver version 0.9.1_beta2 (18/01/2001)
Total: 1 VG 2 PVs 4 LVs (4 LVs open 4 times)
Global: 157993 bytes malloced IOP version: 10 9:21:48 active
VG: raid [2 PV, 4 LV/4 open] PE Size: 4096 KB
Usage [KB/PE]: 38973440 /9515 total 38973440 /9515 used 0 /0 free
PVs: [AA] hda3 18964480 /4630 18964480 /4630
0 /0
[AA] hdb 20008960 /4885 20008960 /4885
0 /0
LVs: [AWDS2 ] user 26214400 /6400 1x open
[AWDS2 ] home 6291456 /1536 1x open
[AWDS2 ] root 5382144 /1314 1x open
[AWDL ] stuff 1085440 /265 1x openOn hdc and hdd I currently have my DVD Drive, and my CD burner.
PS: Please try and keep your answer simple, as I am not very well skilled concerning LVM, allthough I am confident about my general Linux skills.
Quoted 49 lines
If your drives are PVs in one VG, that's not the root system, all you need to do is:
- do a clean shutdown
- change drives
- boot again
- run vgscan (should find your PVs and pack them into one VG) (*)
- run vgchange -a y (should activate your VG) (*)
- mount all LVs in the VGFor me the two steps marked with (*) are done in a initrd and the system would even support changing the root's LV, as long as the /boot stays untouched.
My setup is like this:
/dev/hda1: 64MB /boot/dev/hda2: rest MD0 disk1 (raid 1)
/dev/hdc1: 64MB unused/dev/hdc2: rest MD0 disk2 (raid 1)
/dev/hde1: PV1/dev/hdg1: PV2
/dev/hdi1: PV3/dev/hdk1: PV4
/dev/md0: PV0VG0: PV0
VG1: PV1 PV2
VG2: PV3 PV4hda and hdc are 40GB IBM drives, hd[e-h] were 60GB Maxtor drives.
I'm running a full system out of different LV in VG0 (even root is a LV, only /boot is extra, with inird)
Lately I get a 3ware Escalade controller and moved hd[e-h] there. All I had to do was:
- change initrd to load the 3ware driver (run lilo to install new initrd)
- shutdown, install card, move drives, (remove additional IDE controllers), reboot
- initrd detected all PVs and constructed all my VGs, system still ran fine (you might to have to copy the newly detected configuration from /initrd/etc/lvm to/etc/lvmafter boot)
Quoted 8 lines
(Actually I did it a little different: shutdown, install 6 new 80GB drives, make a RAID5 with this drives, shutdown, add two drives out of VG1, boot (VGs detected on sd[gh]), make PV5 out of RAID5 array, make VG3 with just this PV5, copy all the data from VG1's LV to LVs on VG3, shutdown, remove drives, add drives from VG2, do the same as with VG1, shutdown, remove drives, boot, migration done, system works, all fine.)
Goetz.
----------------------------------------
Content-Type: application/pgp-signature; charset="us-ascii"; name="Anhang: 1"
Content-Transfer-Encoding: 7bit
Content-Description:
----------------------------------------
Quoted 9 lines · Peter Kirk
Quoted 2 lines
Well, as I don't know your current configuration to well, it's just pocking in the dark ...
Here it comes:
I have two disks, currently on hda and hdb.
hda:
10MB of /boot
1GB for SWAP
the lot (18GB) in LV with hdb
Quoted 4 lines
hdb:
19GB in LV with hda[ ... ]
Sorry, but I was unable to come back to you any erlier.
If you still don't have just tried it, I'd suggest you to simply do so :-) (but see the warning at the end).
But only if your /boot contains an initrd, that runs vgscan and vgchange.
If this is the case, you should be able to simply switch the drive to hdc.
IF your system should not boot, because it can't find some LV, don't panik.
In this case you've two options:
- shutdown as clean as possible (CRTL-ALT-DELETE) and switch back. Your system will come up again. (savest version)
- Hope that the boot throws you into a shell (to repair your system), mount the drive that contains /etc as RW (probably mount -o remount,rw /) and do a vgscan (this should fix the /etc/lvm* entries). Now all your LV should be detected again. Now your system should work as before.
Everything basicaly depends on the quality of your initrd and you should be fine if you use the one provided by lvmcreate_initrd.
For savety reasond you should make a backup of the /etc/lvm* files. (or better of all files, but this is most likely no an option)
Goetz.