HPUX 11.23, VxVM and Mirroring to a new array

[markstubb]

Hello:
We're trying to migrate off of our current storage solution to a new array. They are both on the same fabrics and currently the HPUX host can see the disks from both arrays. I'm trying to figure out how I can do this with host mirroring as to avoid downtime on the host. We currently have the base VXVM install, with no additional licenses.

# swlist | grep -i veritas
Base-VXVM B.03.50.IA.014 Base VERITAS Volume Manager Bundle 3.5 for HP-UX

# vxlicense -p
vrts:vxlicense: INFO: No valid license installed

# uname -a
HP-UX xxxxxxx B.11.23 U ia64 2708185549 unlimited-user license

# vxdisk list
DEVICE TYPE DISK GROUP STATUS
c2t0d0 simple - - LVM
c2t1d0s2 simple - - LVM
c8t0d0 simple isdg01 rootdg online
c8t0d1 simple isdg02 rootdg online
c8t0d2 simple isdg03 rootdg online
c20t0d0 simple - - online invalid
c20t0d1 simple disk01 rootdg online
c20t0d2 simple disk02 rootdg online
c20t0d3 simple disk03 rootdg online

# vxprint -ht
Disk group: rootdg

DG NAME NCONFIG NLOG MINORS GROUP-ID
DM NAME DEVICE TYPE PRIVLEN PUBLEN STATE
RV NAME RLINK_CNT KSTATE STATE PRIMARY DATAVOLS SRL
RL NAME RVG KSTATE STATE REM_HOST REM_DG REM_RLNK
V NAME RVG KSTATE STATE LENGTH READPOL PREFPLEX UTYPE
PL NAME VOLUME KSTATE STATE LENGTH LAYOUT NCOL/WID MODE
SD NAME PLEX DISK DISKOFFS LENGTH [COL/]OFF DEVICE MODE
SV NAME PLEX VOLNAME NVOLLAYR LENGTH [COL/]OFF AM/NM MODE
DC NAME PARENTVOL LOGVOL
SP NAME SNAPVOL DCO

dg rootdg default default 0 1128633909.1025.ATLFIN01

dm disk01 c20t0d1 simple 1024 125817280 -
dm disk02 c20t0d2 simple 1024 20966080 -
dm disk03 c20t0d3 simple 1024 134214528 -
dm isdg01 c8t0d0 simple 1024 104856448 -
dm isdg02 c8t0d1 simple 1024 104856448 -
dm isdg03 c8t0d2 simple 1024 20970272 -

v backup - ENABLED ACTIVE 104856448 SELECT - fsgen
pl backup-01 backup ENABLED ACTIVE 104856448 CONCAT - RW
sd isdg02-01 backup-01 isdg02 0 104856448 0 c8t0d1 ENA

v data - ENABLED ACTIVE 125826720 SELECT - fsgen
pl data-01 data ENABLED ACTIVE 125826720 CONCAT - RW
sd isdg01-01 data-01 isdg01 0 104856448 0 c8t0d0 ENA
sd isdg03-01 data-01 isdg03 0 20970272 104856448 c8t0d2 ENA

We're trying to get our data off of the c8xxxx volumes onto the c20xxxx volumes. What license do I need to get from HP to enable mirroring in VxVM and does anyone have any steps they can post on how to do the mirroring?

Thanks!
-Mark

[TTr]

The c8xxxx volumes are members of the rootdg. Do they contain OS filesystems? If so you are talking about server cloning and you should look in IgniteUX.

Otherwise you need to buy the storage foundation product from Veritas or HP to be able to use OS based volume mirroring.

[markstubb]

No, the system install is on local drives. The data and backup volumes are on the SAN storage.

/backup (/dev/vx/dsk/backup ): 47241196 blocks 6298813 i-nodes
/dbms (/dev/vx/dsk/data ): 66414324 blocks 8854341 i-nodes
/opt (/dev/vg00/lvol6 ): 2120000 blocks 33378 i-nodes
/patches (/dev/vg00/lvol9 ): 7499222 blocks 999059 i-nodes
/tmp (/dev/vg00/lvol4 ): 394528 blocks 6183 i-nodes
/usr (/dev/vg00/lvol7 ): 2607072 blocks 41023 i-nodes
/var (/dev/vg00/lvol8 ): 3248784 blocks 51140 i-nodes
/stand (/dev/vg00/lvol1 ): 185936 blocks 2923 i-nodes
/ (/dev/vg00/lvol3 ): 345200 blocks 7313 i-nodes

Thanks for the license info. That should help me get the right license purchased.

Any tips on doing the migration? :-)

[VK2COT]

Hello,

Two simple methods:

a) User:

vxdiskadm

... and select option:

6 Mirror volumes on a disk

or,

b) Use command line with vxmirror
or vxassist. More details at, for example:

http://www.docs.hp.com/en/B7961-90017/ch03s02.html

Regards,

VK2COT

[markstubb]

Thanks, that's what I tried before (diskadm), which failed as I didn't have the license and didn't know it. I didn't know if that was the "right way" to do it.

[TTr]

You have the right approach doing VxVM mirroring but you need the license.

However, looking at your volume sizes you have 125GB of data and 104GB of backups. If you are not using raw volumes for the data, you should be able to do this at filesystem level within a 20minute downtime window (plus database shutdown and startup). Mount the c20 volumes under temp mount points, shutdown the databases, copy the data, flip the mount points (remount the c20 volumes as the original mount points) and start up the databases. This is for the data only, the backups either you repeat the same process or you startup a fresh backup environment on the new volumes.
How fast are your two sets of disks? You should run a test copy of about 20GB to see the performance.

[markstubb]

We were mostly looking at doing a migration via mirroring to avoid the downtime. Currently this server houses a fairly critical application, you know the drill.

The volumes are not raw, so no Oracle or anything here.

The old disk is pretty slow compared to the new disk. Old disk is RAID5 300gb 10k. New disk is RAID1 146 15k, with lots of spindles.

[Khairy]

hi,

i think i know what u mean. I dont have a running hpux w/ vxvm currently but basically
i understands what u require.

u want to mirror current volume to new storage, break the mirror and mount the 2nd copies from new storage?

for that, you would require veritas snapshot and mirroring options.

and u want to migrates disks contents (vol backup & data) to c20xxxx?

ok, u could try the following:

seems the c20x disks has been configured in vxvm.

1) create snapshot for volume backup & data.
# vxassist -g rootdg snapstart backup
# vxassist -g rootdg snapstart data

Continously monitor the status of backup and data volume by using command :
# vxprint -ht

and make sure its syncronized.

The volume will continue to be updated until it is detached.

I'm not sure about your application but if your're running oracle, make sure you alter database to backup mode.

When vxvm complete syncronize, it should show u SNAPDONE state. Now, create snapshot volumes for backup & data. This will break the snapshot mirrors and attach to new volume.
# vxassist -g rootdg snapshot backup snap_backup
# vxassist -g rootdg snapshot data snap_data

Fsck the new volume before mount them.
# fsck -F vxfs /dev/vx/rdsk/snap_backup
# fsck -F vxfs /dev/vx/rdsk/snap_data

Shutdown your apps and mount the mount points to from new volumes.
# umount /backup ; umount /rdbms
# mount /dev/vx/dsk/snap_backup /backup
# mount /dev/vx/dsk/snap_data /data

Hope this helps. Make sure you perform of /backup & /data before proceeding.

Don't forget assign some points if this answers your needs.

Rgds

[markstubb]

So how is mirroring different than using this snapshot approach? Wouldn't mirroring be a better way to go, or does it matter?

[TTr]

>So how is mirroring different than using this snapshot approach? Wouldn't mirroring be a better way to go, or does it matter?

mirroring and snapshots are very similar in the sense that they both give you a copy of your data. Some differences are the mirroring is done at volume level (sectors or extends) while snapshots are done at filesystem level. Another difference is that the snapshots are not complete mirrors but keep track of incremental changes in the filesystem. Typically a snapshot would track the changes in the filesystem and that is what is mounted. It is kind of like an incremental backup vs a full backup. One othe dangerous thing about snapshots is if your original volume changes a lot, the snapshot volume will fill up. I personally prefer volume mirroring but you can use whatever you get comfortable with.