EMC extension

[Mohamed Galal]

dears,
I have HP-UX v11.1. it is connected to EMC LUNs. it is a running system. i need to extend the DB & Backup volumes.
the attached is the current status.
I need to extend /d/db & /d/backup with a new disks.

[Mohamed Galal]

Does it need outage?

[Michal Kapalka (mikap)]

hi,

strings /etc/lvmtab

vgdisplay -v

and i think you need a downtime, but before ignite backup/ data backup.

mikap

[Michal Kapalka (mikap)]

hi,

from you file, there are 2 HBA ( paths )

each path have 13 disk.

mikap

[Mohamed Galal]

Yes I see that i have 13 disks. i will add more disks.
i want to extend /d/db & /d/backup by the new disks.
i need complete procedure encluding outage.

[Michal Kapalka (mikap)]

hi,

i don't know your exact configuration,

but probably you have there a cluster running and this servers is one,

check lvdisplay -v /dev/vgdb/lvol1 |head -n 100

the disk asigned to the vgdb are striped together.

this is not an operation for 10 minutes,

you will need a downtime and exact to know what you are doing, if its a cluster enviroment, you need to make a downtime, aned if the LVOL-s are stiped across disk, you need to probably rectreate the VG-s from scratch and restore the data.

mikap

[Viktor Balogh]

just some questions:

1.) Are these filesystems residing on the luns of the EMC storage?
2.) What volume manager do you use? LVM? Veritas?
3.) Do you have OnlineJFS? What is the output of the following:

# swlist | grep -i JFS

4.) Btw.: What type of FS do you have on the volumes you need to extend?

# fstyp DEVICE
like:
# fstyp /dev/vg00/lvol1
hfs

you can figure out the devices with the bdf command, show me the output of:

# bdf /d/db /d/backup

[Mohamed Galal]

-no, the filesystems residing on the root disk. so, I need to extend the LUNs which carry the Db & Backup.
-LVM is used.
- # swlist | grep -i JFS
gives nothing
see the attached outputs

[Robert-Jan Goossens]

Hi Mohamed,

You have a number of unactivated volumes groups. Do you still need them?

vgdisplay: Volume group not activated.
vgdisplay: Cannot display volume group "/dev/vgnms"
vgdisplay: Volume group not activated.
vgdisplay: Cannot display volume group "/dev/vgosscore".
vgdisplay: Volume group not activated.
vgdisplay: Cannot display volume group "/dev/vgosscore2".
vgdisplay: Volume group not activated.
vgdisplay: Cannot display volume group "/dev/vgmessage".
vgdisplay: Volume group not activated.
vgdisplay: Cannot display volume group "/dev/vgnww".
vgdisplay: Volume group not activated.
vgdisplay: Cannot display volume group "/dev/vgrep".
vgdisplay: Volume group not activated.
vgdisplay: Cannot display volume group "/dev/vgrac".
vgdisplay: Volume group not activated.
vgdisplay: Cannot display volume group "/dev/vgmvi".

Regards,
Robert-Jan

[Mohamed Galal]

these VG are activated on another server in the cluster. so, DB is runnuing here only.

[Viktor Balogh]

>no, the filesystems residing on the root disk

look at the output of bdf:

/dev/vgdb/lvol1 204210176 198326392 5838120 97% /d/db
/dev/vgbackup/lvol1
83656704 434680 82994072 1% /d/backup

vgdb and vgbackup residing on SAN disks, you can figure it out with the help of ioscan and


/dev/vgdb
/dev/dsk/c4t0d1
/dev/dsk/c8t0d1
/dev/dsk/c4t0d2
/dev/dsk/c8t0d2
/dev/dsk/c4t0d3
/dev/dsk/c8t0d3
/dev/dsk/c4t0d4
/dev/dsk/c8t0d4

/dev/vgbackup
/dev/dsk/c4t1d4
/dev/dsk/c8t1d4

e.g. the devices files of vgbackup are pointing to an external SAN disk:

Class I H/W Path Driver S/W State H/W Type Description
==========================================================================
disk 28 0/2/1/0.1.7.0.0.1.4 sdisk CLAIMED DEVICE DGC CX3-40fWDR10
/dev/dsk/c4t1d4 /dev/rdsk/c4t1d4
disk 27 0/6/1/0.1.7.0.0.1.4 sdisk CLAIMED DEVICE DGC CX3-40fWDR10
/dev/dsk/c8t1d4 /dev/rdsk/c8t1d4

[Viktor Balogh]

So, you don't have the OnlineJFS product. Were the new LUNs already presented to the system? After the zoning have been set you should run this to scan for the new devices and install device files for all the newly configured LUNs:

# ioscan -fnC disk;insf -C disk

After this, do a pvcreate for each and every new LUNs, and do a vgextend for both the vgdb and vgbackup.

The next step would be to extend the logical volume with lvextend, as I can see the /dev/vgdb/lvol1 is a simple LV without mirroring, so the PE and LE numbers are equal here.

The last step is to extend the lowest layer, the filesystem itself. You cannot go ahead with fsadm, you would need an OnlineJFS license. Instead you can extend it with extendfs after it has been umounted.
For specific parameters, look for the man pages of each command.

[Mohamed Galal]

Yes you are right, d/db & d/backup not over the root disk.

[Mohamed Galal]

Thank you for your support :)
Can you suppport me with a complete procedure? including the expected outage?

[Viktor Balogh]

You will only have outage in the last 'umount & mount' step, but I don't know how much it takes to extendfs a filesystem. I think it can be done in a few seconds.

[Mohamed Galal]

Thank you,
But can you give me an example

[Viktor Balogh]

you should run this to scan for the new devices and install device files for all the newly configured LUNs:

# ioscan -fnC disk;insf -C disk

After this, figure out all the device files to your new LUNs and do this for each (note that pvcreate works with RAW devices):

# pvcreate /dev/rdsk/cxxtxxdxx

do a vgextend (with block devices, you should present here all the new LUNs you want to add to this FS:)

# vgextend vgdb /dev/dsk/cxxtxxdxx /dev/dsk/cyytyydyy

to extend the logical volume (if you don't have mirroring):

# lvextend -l NEWPENUMBER /dev/vgdb/lvol1 /dev/dsk/cxxtxxdxx /dev/dsk/cyytyydyy

where NEWPENUMBER is the number of PEs you want to assign to this LV. (look in the output of vgdisplay for the freely available PEs)

extending the FS:

# umount /d/db
# extendfs -F vxfs /dev/vgdb/lvol1
# mount /d/db

something like this, I'm not so familiar with extendfs as I solely use fsadm with the OnlineJFS license.

But can I also ask something? Could you assign some points after all of this? They motivate us keeping this forum a helpful source. ;)

[Michal Kapalka (mikap) · Accepted solution]

hi Victor,

the vg's are striped, and if just add aditional disk/disks, you should recreate the LVOL-s from begining, because if you have now in for example VGxyz ==> 6 disk, and stiped cross 6 disks, and if you add 2 disk you need to stripe all from 8 disks, because of performance ....

but this is only my opinion, if you just extend, its not so ok.

mikap

[Mohamed Galal]

ofcourse i am caring about Preformance. and i am caring about no data loss.

[Viktor Balogh]

Hi Michal,

you are right, I didn't notice this.

% lvdisplay /dev/vgdb/lvol1
--- Logical volumes ---
LV Name /dev/vgdb/lvol1
VG Name /dev/vgdb
LV Permission read/write
LV Status available/syncd
Mirror copies 0
Consistency Recovery MWC
Schedule striped
LV Size (Mbytes) 199424
Current LE 3116
Allocated PE 3116
Stripes 4
Stripe Size (Kbytes) 1024
Bad block on
Allocation strict
IO Timeout (Seconds) default