A curse on LABEL=

[Guy Fraser]

Timothy Murphy wrote:

>William Hooper wrote:
>
>  
>
>>>I understand the supposed advantage of LABEL
>>>if one is changing a hard disk,
>>>but I just feel the confusion it causes vastly outweighs its merits.
>>>Am I alone in this?
>>>      
>>>
>>Try adding a new SCSI controller to a system that moves all existing SCSI
>>devices.  You will like LABEL then.
>>    
>>
>
>Surely the proportion of users who want to do this is vanishingly small?
>I imagine < 5% have SCSI main disk, 
>and of these <1% ever want to change the controller.
>(Actually I have a SCSI-only machine;
>but it has never occured to me to change the controller,
>and I would certainly expect trouble if I did.)
>
>But as I said, I understand that the LABEL feature helps
>a very small number of users.
>I'm just saying that it inconveniences a vastly greater number.
>
Give him a break Timothy.

I have never had any difficulty changing controllers on servers with as 
many as 10 drives. I find labels to be of limited use, and usualy modify 
fstab to remove them after I install. But that is my prerogative, I have 
been working with Unix Sytems for 18 years, and I prefer to use fstab 
without labels.

>From what I can tell the only advantage of labels is that you don't have 
to plan a hard drive or partition change as carefully. The main problem 
using labels is that it gives you a false sense that you don't have to plan to move a hard drive or partition, and in some cases it takes extra 
work because you have to relabel a partition before you move it or it 
may conflict when moved.

I currently have an R&D machine with 8 drives and about 15 partitions.
On that machine labels have been both a blessing and a curse at 
different times when moving data and mount points around. I could live 
without them, but they are likely here to stay, so we will just have to 
get used to them.