[linux-lvm] potential bug with LVM2, --removemissing, and a LVM1 VG

[Peter Smith]

I'm posting this to the list before I enter a real bug report 
just-to-see if anyone can see the flaw in what I'm doing...

First the background.  Recently I removed some SAN space from one of our 
system's LVM1 setups.  Problem is, I did a 'pvchange -x n <pv>', a 
'lvremove <lv>'  on the SAN space in question and then basically removed 
the space from the system physically.  Well, the rest of the VG 
obviously refused to mount after a reboot..  Through many steps I 
learned that the '--removemissing' option for 'vgreduce' on the newest 
LVM2 is bugged in that it will not work with LVM1 VGs.  You are forced 
to use a non-bugged version of LVM2.  As a matter of fact, you can 
create a LVM1 VG under LVM2, perform the steps above, and then try to 
fix it using the '--removemissing' option and it will NOT work.  I have 
gone so far as to create a script that can exercise the functionality 
and illustrate its failure.

So my question is, is this a legitimate bug?  I believe it is.  
Specifically a bug with the '--removemissing' option for the newest LVM2..

Thanks,
Peter