creating a logical volumen on specified physical volume?

[golfnut]

Is this possible.

i have a vg with multiple pv. I want to create a lv on a specific disk.
And then mirror to another specific disk.

I was looking at lvcreate and looks like it does not have an option like that.
I am thinking there must be a way to do this.

Obviously new to HP-UX.

thanks

[Bill Hassell · Accepted solution]

Generally, lvols can be laid out without regard to the PV but there is a way to do this. Create the lvol without any length. Then use lvextend to extend the zero-length to the size you want and specify the PV path you want to use. Then use lvextend -m 1 and specify the PV for the mirror extents.

[Devender Khatana]

Hi,
For your convience commands you need to follow :-
I am assuming your disk /dev/dsk/cztzdz is allready added to your VG /dev/vg0x And all Logical extents of disk are free. Verify it by pvdisplay command
#lvcreate /dev/vg0x
#lvextend -L ??? /dev/vg0x/lvolx /dev/dsk/cztzdz

Precaution should be taken while giving Size in MB's. It should not be more than Physical disk size otherwise it throughs an error.

#newfs -F ???? -o ???? /dev/dsk/lvolx

Put a entry in /etc/fstab

HTH,
Devender

[bhavin asokan]

The following commands create a 500MB large LV named lvdata on any disk(s) of the VG
vg01:
# lvcreate -n lvdata -L 500 vg01

You cannot specify a PV with lvcreate. If you like to put the LV on a certain PV use the
following:
# lvcreate -n lvdata vg01
this creates a LV of 0MB. It has no extents - it just exists. Now extend the LV onto a certain
disk:
# lvextend -L 500 /dev/vg01/lvdata /dev/dsk/c4t2d0
Now you can use newfs to put a FS onto the LV:
# newfs -F /dev/vg01/rlvdata

regds,

[golfnut]

Thanks for your respone.

What is the best way of copying filesystems in HP-UX?

I am going to be creating a new usr an new var file systems. I want to then copy the usr and var to these new file systems and then make the change in fstab.

thanks

[Bill Hassell]

Just for clarification, here are the steps:

1. Pick your VG, in this example I'll use vg01. Pick the disk path you want to use. To see what is available in the VG, use: vgdisplay -v vg01 which shows the physical disk paths at the end. For mirroring you should have a minimum of two disks in the VG. We'll use /dev/dsk/c2d6t0 as the target disk and c2d5t0 as the mirror:

2. lvcreate vg01
(That's all you need. The next unused lvol name will be assigned automatically and the size will be zero. We'll assume that lvcreate made the lvol: /dev/vg01/lvol3

3. lvextend -L 500 /dev/vg01/lvol3 /dev/dsk/c2d6t0

4. Now the lvol exists with a size of 500 megs on the c2d6d0 disk. Since you want to mirror this disk, you can do that now before you create a filesystem on it. It doesn't matter which comes first, mirror or create filesystem. Specify a different disk for the mirror:

lvextend -m 1 /dev/vg01/lvol3 /dev/dsk/c2d5t0

If you can see the disk lights, you'll see them both flashing as the lvol is synced.

5. To see what your lvol looks like, you can run lvdisplay -v /dev/vg01/lvol3 after step 3 and then step 4. You'll see each extent and where the extent resides.

6. Run newfs to create a filesystem. The default should be VxFS but just to be sure, specify the filesystem type:

newfs -F vxfs -o largefiles /dev/vg01/rlvol3

NOTE: LVM commands can refer to the lvol or rlvol device but newfs must always use the raw (character mode) device file which is the lvol name with r in front. The -o largefiles will allow the creation of files larger than 2Gb.

7. Create a mount point: mkdir /mnt1 (or whatever you'd like to call it) then edit /etc/fstab to add the new mountpoint:

/dev/vg01/lvol3 /mnt1 vxfs nosuid,largefiles,rw,delaylog 0 3

nosuid is recommended for all mountpoints except specific HP-UX mountpoints. largefiles is needed to enable the largefile support on the lvol. rw and delaylog are standard options.

8. Now test the fstab entry by mounting the mountpoint: mount /mnt1

9. Verify that it exists: bdf /mnt1

And that's it.

[golfnut]

thanks for the reponse guys.

All the exisiting file systems on this server are hfs. I dont know why?

Are there any issues moving them to vxfs?

uname -a give the following.
HP-UX ux01pwow B.11.00 A 9000/800 658349303 two-user license

thanks

[Fabio Ettore]

Hi,

check that your system has JFS installed and configured.
For this launch:

swlist -l fileset -a state | grep -i jfs

Up today I cannot find systems with all hfs filesystems, your is for museum :-)

I don't see problems to migrate them to vxfs. Of course you should backup your system.


For example a your hfs filesystem is test on logical volume /dev/vgXX/lvol1, then:

umount /test
newfs -F vxfs /dev/vgXX/rlvol1
fsck -F vxfs /dev/vgXX/lvol1
mount /dev/vgXX/lvol1 /test

restore from a your previous backup.

By the way if all your filesystems of system are hfs then I should re-install completely your system, I thonk it is for more safe.

Mind to have only /stand filesystem in hfs type.

Best regards,
Fabio

[Steve White_7]

When creating logical volumes you should use the default names. If the default names are not used and you ever have to vgimport, you are not going to have a nice time on a big system.

[Bill Hassell]

HFS is a good filesystem but has limitations on maximum filesystem and file size (128Gb), and in case you accidently lose power or you have a system crash, the fsck time is VERY long for multi-gigabyte filesystems (dozens of minutes). I would not convert existing HFS filesystems to VxFS on a production machine...HFS is quite stable, although it is a bit slower than VxFS for directory and disk space changes.

he safe and fastest way to convert everything (except /stand which must always be HFS) is to use Ignite/UX. Download the latest version from http://software.hp.com and run make_tape_recovery to backup vg00. I'd also make an fbackup copy of your entire system. Then restore the Ignite/UX image but modify the disk layout to VxFS. At the same time, you can resize any of the lvols.