On 3/2/12 9:13 AM, Jesse Butler wrote:
Hi Drew-
I've got a wrinkle here, and I'm sorry as it seems to be related to a
failure of imagination on my part.
In looking at your screens, I am realizing that the user will be
required to specify a LUN number, and that only that particular iSCSI
LUN will be presented to them for installation. I believe that's not
correct. I understand that we specify a LUN number based upon my input
regarding AI manifest configuration for iSCSI targets. That is because
we can't guess at what levels of LUN masking are in play on the
initiator, and we don't just want to arbitrarily start installing to
iSCSI LUNs that "look right"... so I suggested that we require a LUN
number specification. Now, we come to my failure of imagination...
I had not at the time understood that the decision to require LUN
number would play forward into usage in the interactive installers. I
would think that users should be able to specify an IP address, IQN,
etc and have the LUNs available to them presented, just as is the case
with other disks.
A good way to look at this is to compare the iSCSI configuration to an
array... a iSCSI target (IP addr, IQN, etc) is the array, and the LUNs
are the disks inside. If you have a Fibre Channel array attached to a
machine, you would list all of the disks in the install screens for
selection, right? I think the same behavior would be expected with iSCSI.
Let me know if you want to discuss this. If the upshot is that it's
too much to consider at this point, I'd be fine with specifying the
LUN now, and having an RFE for the future.
You're absolutely right and QE has pointed this issue out as well. I'll
investigate making the target LUN optional. If not provided, return all
of the LUNs from the target. If it is provided, display only that
specific LUN.
How does that sound?
-Drew
_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
