Leadville driver and persistent binding

I guess the leadville vs. 3rd-party drivers (and their configuration thereof) can be baffling for those of us who haven't used leadville driver (and sun's oem HBAs) in the past.

After presenting the LUNs to the hba, use a combination of cfgadm, luxadm and devfsadm to discover and configure your LUNs.

This is absolutely my opinion, but the way the leadville driver addresses the disks/LUNs with the WWN of the LUN as the target number (resulting in a huge cntndn path -- eg: c1t5000<blah...blah>AEDEd0) seems completely creepy.

To those who have gone before me --

Is there any way to prevent the leadville driver from doing that and present a more easily readable, parseable and script-friendly naming scheme?

implicate_order at 2007-7-6 21:27:16 >

if you pkgchk - do you see a SUNWqlc package? I'm not sure if your specific HBA is covered by that package, but from what I remember the support folk saying, you need to have the Sun branded Qlogic cards in order to be able to use the leadville drivers.

Also, you may want to check the download portion of sun.com for the SUNWsan package -

If you do format < /dev/null and look at your traditional CTD's, you'll be able to tell if you have leadville drivers installed or not as the WWN will now be incorporated into the device name (at least it's this way on 9, my guess is that it's this way on 10 as well)

here's a for instance of a format entry with leadville

6. c2t50061485C700FE98d00 <EMC-SYMMETRIX-5670 cyl 36826 alt 2 hd 30 sec 64>

/pci@8,600000/fibre-channel@1/fp@0,0/ssd@w50061485c708fe98,34

jeffreys at 2007-7-6 21:27:16 >

> This is absolutely my opinion, but the way the

> leadville driver addresses the disks/LUNs with the

> WWN of the LUN as the target number (resulting in a

> huge cntndn path -- eg: c1t5000<blah...blah>AEDEd0)

> seems completely creepy.

>

> To those who have gone before me --

>

> Is there any way to prevent the leadville driver from

> doing that and present a more easily readable,

> parseable and script-friendly naming scheme?

No. You can use MPxIO/STMS to get an even longer and creepier name. :-)

While it's a little strange at first (and more difficult to type), I'm not sure why having a WWN as the target is any less parsable or script-friendly.

--

Darren

Darren_Dunham at 2007-7-6 21:27:16 >