Remove a file system without removing the underlying logical volume

[columb]

Is this posible.

Our system involves a complex SAN backend and creating logical volumes is a pain in the fundament. A file system was created incorrectly (not large file enabled) and needs to ge rebuilt. If I run rmfs I remove the logical volume and have to recreate it. Any ideas?

Ceci n'est pas un signature
Columb Healy

 

[mrn]

You could create another small filesystem on the lv the rmfs would then not remove the lv

Mike

"A foolproof method for sculpting an elephant: first, get a huge block of marble, then you chip away everything that doesn't look like an elephant."

http://www.airport-parking-site.co.uk/

 

[p5wizard]

I edit /etc/filesystems and remove the FS's stanza. (unount the FS first).

Then create the FS anew.


HTH,

p5wizard

 

[Breslau]

i don't understand the question. an rmfs will not remove the LUNs (hdisks). even if you're striping accross multiple disks, you just reselect the ones you need upon creation.

even if lv creation step is somehow complicated, if you used smit to make it, just go back and rerun the command, assuming you haven't nuked root's smit log since then.

 

[p5wizard]

Columb is talking about a logical volume (compare it to a disk partition) in an AIX volume group. LVM stuff. Nothing to do with SAN LUNs or SAN volumes.


HTH,

p5wizard

 

[columb]

We have two regatta systems in two data centres each with it's associates shark disk cabinet.
The two sharks are mirrors, and the two regatta systems are a number of clustered pairs.
The OS is AIX 5.1 ML7
I don't realy understand HACMP in more than the most general terms but the HACMP guru is off today and I'm trying to sort out a file system that was built without large files being enabled. I do know that if you simply use smitty then it can create file systems where the two halves of the mirrors are in the same cabinet - or something, I just know that the HACMP guru insists that we don't use smitty.
mrn - I can't create another file system, the lv is full. It's a good idea though.
p5wizard - I'm not sure that I want to hack /etc/filesystems on one of our main production boxes. If I got it wrong then I'd be looking for another job.
Looks like I'll have to wait until tomorrow when the HACMP guru is back.
Thanks for your input.

Ceci n'est pas un signature
Columb Healy

 

[Breslau]

ahh ok. then i'd just expand on what others have already said.

you could create a small lv in your rootvg and change the mount point of the filesystem with chfs:

chfs -a dev=/dev/[newLV] /[filesystem]

this way you would not have to touch the filesystems file directly.

then run the rmfs as you normally would. this would get the stanza out of filesystems, again w/o touching it directly.

re-create the filesystem the way you want.

make a backup copy of /etc/filesystems beforehand for extra paranoia

 

[p5wizard]

Well, if you delete a stanza for a FS (6 or 7 lines beginning with the mountpoint pathname followed by a colon) in /etc/filesystems, it doesn't touch the data at all, doesn't even unmount anything, just means you can't automatically mount it again on system reboot or with "mount all". You just have to make sure you get the right stanza... But if you don't want to edit the file, that's fine with me.

If you have a spare disk out of the HACMP realm or a testvg on a testserver you can try it out there. Otherwise, you wait...


HTH,

p5wizard