| Thanks guys - I can now report some further information - FCP connected tapes, unlike FCP connected DASD, are not usable under z/VM. I understand they are defined to the system, but as of 5.2 - the code necessary to make them available is just not there. (That is a $140K bite out of my rear end... *sigh* )
However, they are available to Linux guests - at least the documentation says they are. There seems to be a totally arcane methodology of connecting them within Linux that I will be decoding over the next couple days. If anyone is interested, I'll write up a few paragraphs and post it once I get the so-and-so things to talk to each other.
Heck, this is getting critical here due to some time frames that got seriously compressed by delayed equipment deliveries. If anyone is in Austin and wants to take a crack at 'em - let me know. I suspect I will be at work all weekend... :)
-Paul
On May 8, 2006, at 6:03 PM, Eric R Farman wrote:
Hi Paul,
When issuing a "SET EDEV xxxx TYPE FBA ATTR yyyy [paths]" command, the "xxxx" value should be an unused RDEV number that will be created. The FCP RDEV number (D200) is used later in the command, following the "FCP_DEV" operand.
Hope this helps you avoid your maintenance window.
Regards,
Eric
Eric Farman
z/VM I/O Development
IBM Endicott, NY
(607)429-4958 (tie 620)
Steve Wilkin's page is a goldmine! Thanks for the pointer. :)
I see I gotten myself in a conumdrum here - the FCP port if available, but I should not have defined any devices on it.
"SET EDEVICE xxxx xxx xx xx xx ... " results in a complaint that 'D200 is a defined as a real device". I suppose it is
time for yet another maintenance window, and bring her down again. (*sigh*)
I should not be this difficult. :) :) :) :)
I suppose I am going to have to start a VM User group down here - the closest one I know of the CaveMen guys up in Chigaco. :)
-Paul
----- Message from "Alan Altmark" <[EMAIL PROTECTED]> on Mon, 8 May 2006 20:27:00 +0000 -----
On Monday, 05/08/2006 at 04:20 GMT, Paul Raulerson <[EMAIL PROTECTED]>
wrote:
> I modified the IODC (using the HMC, I have not figured out how to use
the z/VM
> software for doing so yet... :) to change a free FC channel to a FCP
channel,
> and added a CUNUMBR line to support. I?ve included the definition file
below,
> with the interesting bits marked in red.:)
>
> From z/VM I can do a query on the CHPID (F3 in this case) and I see that
it
> thinks there are devices available out there. That is kind strange,
because the
> new tapes are not yet attached there are two new 3592 TS1120 tape
drives ready
> to connect to the SAN switch and configure. The FiCON port is also ready
to
> connect to the SAN switch and, if necessary, do any configuration.
Actually, VM doesn't see the SCSI devices. What it sees are the
subchannels on the chpid. In normal ESCON attachment, the control units
actively participate in the I/O process. They know which devices are
available and which devices are not.
For SCSI, there is no traditional control unit to manage device attachment
to the I/O subsystem. There's just a cable the plugs into your FC switch.
The connection to a particular drive is established dynamically via a
specific protocol between the guest or CP and the switch.
So as long as the chpid is active, all of the devices you defined on the
chpid will appear, but they don't have active devices on them until CP
(via EDEVICE) or a guest (via its own SCSI device drivers) does something.
> Problem is, from this point on I am lost there are WorldWidePort
addresses and
> a lot of other terminology I am not familiar with. So any references or
> pointers would be very much appreciated. I have a feeling that this is
not all
> that hard. The Shark is current attached to the same SAN switches, and
provides
> DASD for the PC?s.
See Steve Wilkins' home page: http://www.vm.ibm.com/devpages/wilkinss
where you will find several of his presentations on z/VM and SCSI. These
are excellent resources. (Steve is the Father of z/VM SCSI.) It helps if
you can sit down with someone who has storage expertise to help round out
some of the practical considerations.
And get thee to a VM user group near thee! If the user group requests
someone come and speak on SCSI, we are more than happy to oblige.
Alan Altmark
z/VM Development
IBM Endicott
|
