Solaris

On think to note is that this assumes that the arrays are configured with the same number of LUNs.

Obviously given the amount of information in this I can't suppy you will all of the command to do all of these tasks in one post, but it should get you pointed in the right direction.

1. Stop any databases you have running on your system.
2. Detach the mirrors of each of your volumes
3. Dump the inactive mirrors of your data volumes to tape
4. Shudown and remove the 2nd mirror of the root disk or dump it to tape
~~~then on the opposite end
5. Detach mirrors
6. Restore root disk to detached mirror of the root disk (or replace it with the disk you removed from the first machine)
7. Create data volumes on the detached mirror and restore from tape
8 Reboot to single user mode using the (currrently detached) 2nd mirror
9.Apply the new veritas lincense & reboot again using the 2nd mirror
10. If every thing is working remirror using the 2nd mirror as the master

Note that using this approach there is no downtime on the target system until #8

Another approach would be to use Veritas NetBackup to achieve most of this.

sry for reading that post late,

but:

with veritas you bought a very good developed volumemanager, there is no need to detach mirrors or volumes in VxVM! if you really have the same hardware, you only need to put the "old" things in the new one.

in a v240 there is a card which is called the "System Configuration Card" (SCC) where the HOSTID is stored (hostid and nvram settings). you need to put the old card in the new mashine to provide the right HOSTID to your VxVM license.

in theory you could put in those disk in any order... that'S a feature of VxVM... when the kernel starts a demon (vxconfigd) looks at all attatched disks and reads the privat region on the disk to see which diskgroup and which volume (data in public region) configuration is on the disk [manual command would be: "vxdctl enable"] that's a big different to SDS/LVM because it handles the volumes with local databases (metadb) but VxVM stores the information on each disk. that's the reason for the logical split of the disk into privat and public regions.
###
by the way: there are also feature for deporting and importing. take a look at the command vxdg {deport|import]. to see all diskgroups not only the imported use that command: "vxdisk -o alldgs list"
###
the dual way to the disks should be enable afterwards with the veritas DMP. enable it for the disks. (could be done with the "vxdiskadm" cli interface or with the cli command "vxdmpadm") i'm not sure if DMP will work if there are no controllers in the 3310 just a jbot. take a look at the instructions in the "Enabling VERITAS DMP" section of the Sun StorEdge 3000 Family Installation, Operation, and Service Manual on docs.sun.com.

anyway. the idea with the backup should not be ignored for safety reason ;-) but it should work with the normal disk&SCC replacement..

greetings pressy

All true Pressy, with the following qualifiers:

True, but in that case why not just bring the whole hardware hardware setup? That too would be easier. There may also be licensing issuse here. As a Veritas (and Sun) OEM supplier I know that we cannot do this when installing for external customers.


In theory yes, in practice with the 3310 you cannot, unless it is a JBOD. In which case it can be done this way. It cannot be done in a cotrolled bases 3310 as a global replacement of disks would require rebuilding of the logical drives, wiping all the information, also insering the disks in any order would not work in the 3310 as the logical drives are configures based on the physical position of the disks in the array.

This is what I use and why I at first didn't consider the JBODs, we only use them as for expansion on controller based 3310s. And the reason I said a direct swap probably wouldn't work.

In single bus mode DMP configuration can be done on a JBOD, but since he is using dual bus it's not an option.

It also give hardware independence, you could happily move from a 3510 to any other array by just adding the drivers.

VirginiaSA, read Pressy's reply in detail, it will work if you are using JBODs with no controller. On the other hand, if you really want to take the physical swap approach, it might be easier just shipping he whole lot to where it's going and swap the two systems.

Well I guess pressy is right. come to find out, it is a JODB and single controller mode. 3310 has one controller for each side of the chassie.

I'm more confised now after this.

Single bus or single controller? If it has a controller it is not a JBOD, and pressy's way will not work. A single controller 3310 is simply that, one controller for the entire array using two SCSI channels. If there are two controllers on the array than it is dual controller using four SCSI channels.

Basically which of these does the back of it look like, and for the controllers if it is the top one it looks like, the simplest is how many com ports does it have?