| 1 |
On 19/12/2020 17:32, Dale wrote: |
| 2 |
> Neil Bothwick wrote: |
| 3 |
>> On Sat, 19 Dec 2020 07:02:23 -0600, Dale wrote: |
| 4 |
>> |
| 5 |
>>> I have another question related to LVM. Let's say a system crashes and |
| 6 |
>>> dies. Or I just move a drive, or drives, with LVM on it to another |
| 7 |
>>> system. Does the system just recognize the drives and knows how to add |
| 8 |
>>> them or do I have to do that manually on the new system? |
| 9 |
>> It should just work if you move all the PVs in the volume group. One |
| 10 |
>> thing to watch out for is if the destination system is already using LVM, |
| 11 |
>> the volume groups must be named differently. That's why I always use |
| 12 |
>> unique VG names based on the hostname. |
| 13 |
>> |
| 14 |
> |
| 15 |
> Ahhh, so it stores the info on the drive so that it knows what it is. |
| 16 |
> Neato!! I was thinking it was in /etc/lvm/ or something. I've wondered |
| 17 |
> about that for a while now. |
| 18 |
|
| 19 |
mdadm version 0 stored its information in mdadm.conf. That was a mistake |
| 20 |
- it had the downside you couldn't boot from the array, it was wide open |
| 21 |
to errors, arrays were regularly trashed because things had got |
| 22 |
confused, etc etc. Sticking all the necessary information in a |
| 23 |
superblock is now considered must-do good practice. |
| 24 |
> |
| 25 |
> Naming them based on hostname is also a good idea. I don't have mine |
| 26 |
> named that way but if I ever redo them, I will. It could save me some |
| 27 |
> problems down the road. |
| 28 |
> |
| 29 |
Again, that's now the default for mdadm - not necessarily the user name, |
| 30 |
but the internal array name is something like "tigger:0", to quote one |
| 31 |
of mine - array 0 created on tigger. |
| 32 |
|
| 33 |
Cheers, |
| 34 |
Wol |
Archives