Took a look at the Primus that you referenced. Basically here is what it is telling you the steps are.
Write disable the device (at this point you would still be able to read it, just not make any writes to it) Get list of where this device is mapped to (will be used as a reference later) Unmap the device from the FA's Make your changes to expand the META Verify your changes Remap the device back to the FA's Write enable the device
It's been a while since I've explanded a CONCAT meta device. Last time it was on a Symm5.5 running 5568 code.
Now, you CAN expland a STRIPED meta online, using a BCV to preserve the data.
BUT (if I recall correctly) you CANNOT expand a CONCAT meta online. It must be unmapped (although you may get away with just making it not_ready - which is far easier to achieve BTW). I'm not sure why this restriction exists.
You'll therefore need to plan an offline slot for the host.
The data will be preserved, but you'll also need to "expand" the disk from the host OS as mentioned above.
Looks like it's better to convert your beloved meta to striped and then expand it while preserving your data .. Since both operations are online, if your device is local (no SRDF involved) you can build a BCV and use it while converting your meta to striped and while expanding it ..
Did this behaviour change at some point? I am fairly certain that I have expanded concatenated meta devices in the past without taking an outage and I doubt I was just lucky that it worked
In fact that was one of the advantages being touted for using concatenated metas many years ago over striped metas from what I recall. Of course now we are pretty much striped metas across the board so can't easily test now. Checking online ECC and symconfigure documentation and is not very clear if this is supposed to be an online operation or not. ECC describes the striped expansion in detail.
Did this behaviour change at some point? I am fairly certain that I have expanded concatenated meta devices in the past without taking an outage and I doubt I was just lucky that it worked
that´s what i thought : the reason for using concat metas is only to expand them in an easy way online!! And when u do it im Mcode it looks like an online config change! But this symmcli primus describes it as offline for host OS
I'm pretty curious since I used to tell in my Symmetrix Configuration classes that concat meta can be expanded online .. However recent tests ran from other guys pointed toward the opposite direction.. But your experience confirm that YES WE CAN !! :-D
Dynamox can you please tell us Solution Enabler version and code running in the box ?? I'm pretty curious since I asked a collegue to run some tests and he confirmed that -in his very specific environment- a concat meta can be expanded only "offline" .. Unfortunatly I can't run tests in my usual "test bed"
MikeMac1
2 Intern
•
292 Posts
586
0
Posted June 9th, 2008 08:00
Write disable the device (at this point you would still be able to read it, just not make any writes to it)
Get list of where this device is mapped to (will be used as a reference later)
Unmap the device from the FA's
Make your changes to expand the META
Verify your changes
Remap the device back to the FA's
Write enable the device
Let me know if this answers your question.