Adding drives to rp5450

[Ian Hillier]

I need to extend some file systems that are on vg00 and I bought 2 more drives and plugged them in. vg00 is mirrored. My question is - should I import these new drives into vg00 and grow the fs or should I create a new vg with these drives, make new (mirrored) file systems and then extend the existing across the new? My thought is to do the latter but I'd like some confirmation.

Thanks!

[RAC_1]

Depends on your requirement.
If your file sytem growth is high I would suggest you to go for option1 ese option2.

Regards,

[Helen French]

Some points:

1) You cannot IMPORT a new disk ! vgimport will work only if the disk has some old VG information on that.

2) You cannot extend the LVs if the filesystems are on different VGs.

Two options to try:

1) Create new VG on the new disks, create file systems and mirror it. Create links on the vg00 which points to the new file systems.

2) Add these disks in vg00, extend the file systems you need and mirror it.

[Stefan Farrelly]

You will only be able to add these new drives to VG00 if they are the same size or smaller than your current drives.

If you are going to use them for VG00 lvols (/opt, /var, /usr) then add these disks to VG00. If you are going to use them for applications (eg. /apps, /db etc.) then normal procedure is to put them into a new VG (VG01)

[Jeff Schussele]

Hi John,

IF you want to extend vg00 then you must vgextend vg00 - note not import. Then lvextend the /dev/vg00/lv_volname.
You'll want to
pvcreate /dev/rdsk/cxtydz (both drives)
vgextend /dev/vg00 /dev/dsk/disk1 /dev/dsk/disk2
lvextend -L xxxx /dev/vg00/lv_volname (for all needed)
Check the LVs w/lvdisplay to insure that the mirror is intact for the new devices.

But my question would be why are you extending vg00 - is it app space you need? If so I'd look at creating a separate VG for the apps & linking it to wherever it was before. It's always better to keep apps separate from vg00.
If it's /var for log sizes - then keep the logs trimmed - archiving if necessary. If it's users who need BIG home dirs - you could create another VG/LV/FS & move the existing user data to it, remove the existing /home & mount the new VG/LV/FS to /home
If you HAVE to extend vg00 you very well may need to do this in single-user mode.

You cannot extend an FS from one VG/LV to another VG/LV. You *could* link an FS from another VG/LV into the dir structure of an FS in an existing VG/LV

HTH,
Jeff

[James R. Ferguson]

Hi John:

I'd avoid adding more physical disk to vg00. It is much easier, from a maintenance standpoint, to deal with a single physical disk for the boot volume (plus its mirror, of course.

BTW, you would 'vgextend' the new disk into your existing vg00 (you don't 'vgimport' them, since there virgin disk) if, however, you feel you want to proceed in the fashion you are thinking.

*However*, if your original vg00 volume group was created with default 'vgcreate' parameters, and a small disk with limited physical extents (e.g. 4GB worth) then you will *not* achieve full utilization of your new, larger disks (e.g. 9GB or larger) since the number of supported physical extents is fixed at 'vgcreate' time (see the man pages for more information).

If you have non-standard filesystems as a part of vg00 (i.e. other than '/', 'usr/', '/opt/', 'var/', 'tmp/', etc. then give strong consideration to moving them to a (new) volume group other than vg00.

If you have a small, vg00 and your new (physical) disk is large (e.g. 18 or 36GB), then I'd make an Ignite 'make_tape_recovery' tape and reinstall using the Ignite advanced installation tab to resize your filesystems, using only *one* physical disk.

Regards!

...JRF...

[Shahul]

Hi

U can not use import. There are two options open for U

1)Add these two new disks to vg00

2) Create a new vg and new FS

It is better to go for IInd. Because adding more number of HDDs to vg00 is not suggested.

Follow this

#ioscan -fnC disk

Note down the device files of newly added HDDs

Confirm those by using
#diskinfo /dev/rdsk/cxtxdx
then
#pvcreate -f /dev/rdsk/cxtxdx Do it for both
#mkdir /dev/vgxx
#mknod /dev/vgxx/group c 64 oxab0000

here "ab" in 0xab0000 should be the vg's number. Then
#vgcreate -e 18000 /dev/vgxx /dev/dsk/cxtxdx
#vgextend /dev/vgxx /dev/dsk/cxtxdx

#lvcreate -L /dev/vgxx
Here -m 1should be added if U want mirroring

Same do for other lvols also.

#newfs -F vxfs /dev/vgxx/rlvol

Then mount these lvols as per ur requirments. After mounting U can copy data from vg00 to this file system, Or U can copy this and link to vg00. One more thing U can do like, Mount this FS to one of the subdirectory of the old file system which U want to extend.

Best of luck
Shahul

[S.K. Chan]

You CANNOT extend an existing LV in vg00 across to a different VG (say vg01). To best access your situation please provide output for these ..
# vgdisplay -v vg00
# bdf
Generally vg00 should only contain "system" LVs (eg : /opt, /usr, /var). If you have "data" LVs in vg00, you should migrate them out to a new VG. If you do not have any "data" LVs in vg00 and simply want to extend its LVs then you would have to add these 2 disks to vg00. That'll make things a bit complicated now that you got to think about how to mirror those disk properly by making sure 2 sets of disks are identical (using PVGs).

[Ian Hillier]

Sorry for the confusion, I used the term import loosely as I'm using SAM to do this. SAM sees the new disks, but ioscan does not, even after insf.
Seems I have some migrating to do... YUK!

Thanks to all...

[Sandip Ghosh]

You have to give insf -e to create the device files for the new disks.

Sandip

[Sandip Ghosh]

Regarding the addition of the disk , I would create a new volume group and and those two new disk over there. After creating the new lvol/filesystem on the new disk, I would copied the existing data of the filesystem which needs to grow to the new filesystem. This is basically to keep the system clean.

Sandip

[Ian Hillier]

That's the plan. New vg with new drives, bigger (mirrored) lvols, mount to a temp, move data, umount old, mount new, move on, lesson learned.