Opensolaris disk names using LSI HBA?

D

danswartz

2[H]4U
Joined
Feb 25, 2011
Messages
3,715
So I made the cutover - the M1015 flashed with IT firmware is now running my 6 sata disks. I noted when the zpool was imported, the device names are funky. Like this:

NAME STATE READ WRITE CKSUM
tank ONLINE 0 0 0
mirror-0 ONLINE 0 0 0
c5t50014EE2AEDF73CEd0 ONLINE 0 0 0
c5t50014EE2AF872299d0 ONLINE 0 0 0
mirror-1 ONLINE 0 0 0
c5t50014EE2AEDF7498d0 ONLINE 0 0 0
c5t50014EE0AC01D8EDd0 ONLINE 0 0 0
mirror-2 ONLINE 0 0 0
c5t50014EE204411A53d0 ONLINE 0 0 0
c5t50014EE0ABCEE0A9d0 ONLINE 0 0 0


Any idea what the funky things are after the c5t5 parts? Usually, opensolaris gives you something like c2t2d0s0 or somesuch...
 
Strange, almost looks like it imported them by a partition identifier or something. I've had a similar issue on ZFS on FreeBSD where it imports a pool by GUID's but that's not whats happening in your case. Probably nothing to worry about, just makes it more difficult to figure out which logical drive equates to a physical drive.
 
I'm tempted to reboot and go into the LSI HBA BIOS and see if it is describing the disks some such way.
 
Might also check to see if the drives serial number is part of the name, or any other drive attributes that match the dev ID's, I've never seen Solaris name things in such a funky way.
 
With IDE/ Sata/ SAS1 Controller, you have port numbers like c0t0d0
They refer to a physical port of your controller (ex first port on contrioller c0)
If you move such a disk to another port, it gets a new number
(does not matter because config is stored on disk)

LSI changed this behaviour with their SAS2 Controller. The now use disk-id (WWN numbers)
numbers like c5t50014EE2AF872299d0
These numbers are unique to a disk. If you plugin such a disk into another slot
it keeps this number.

How to deal with:
When you insert a disk first time, you must write down the number
In case of problems, you must know the slot, where this disk is inside
(you may try the disk identify via dd function in napp-it to identify, works not in any case)

If you want more infos, google about WWN and read
[ame="http://en.wikipedia.org/wiki/World_Wide_Name"]World Wide Name - Wikipedia, the free encyclopedia@@AMEPARAM@@/wiki/File:Crystal_Clear_app_network.png" class="image"><img alt="Stub icon" src="http://upload.wikimedia.org/wikipedia/commons/thumb/4/49/Crystal_Clear_app_network.png/35px-Crystal_Clear_app_network.png"@@AMEPARAM@@commons/thumb/4/49/Crystal_Clear_app_network.png/35px-Crystal_Clear_app_network.png[/ame]
 
if you use WD drives they have WWN printed on their label, otherwise you have a keep a list with sn - wwn association.
 
An interesting thing to note for the M1015 in particular, is with the normal firmware with the LSI driver it uses the "old" drive naming style. If you cross flash it to a 9210 though it uses the WWN naming.
 
apnar said:
An interesting thing to note for the M1015 in particular, is with the normal firmware with the LSI driver it uses the "old" drive naming style. If you cross flash it to a 9210 though it uses the WWN naming.
Click to expand...

The same with me
My SuperMicro X8 boards with integrated LSI 2008 controller flashed to IT mode uses WWN
while my new LSI 9211 controller uses short port-id's out of the box.
Undecided if I try to reflash to something other because short port numbers are mostly
more practical than WWN's
 
Now I am really confused. So I have two spare ports on my M1015 (IT firmware). I got out an old 80GB sata drive (WD) and hot-plugged it. I now have this:

0. c2t0d0 <VMware -Virtual disk -1.0 cyl 2085 alt 2 hd 255 sec 63>
/pci@0,0/pci15ad,1976@10/sd@0,0
1. c5t50014EE0ABCEE0A9d0 <ATA-WDC WD6400AAKS-0-4C07-596.17GB>
/scsi_vhci/disk@g50014ee0abcee0a9
2. c5t50014EE0AC01D8EDd0 <ATA-WDC WD6400AAKS-0-4C07-596.17GB>
/scsi_vhci/disk@g50014ee0ac01d8ed
3. c5t50014EE2AEDF73CEd0 <ATA-WDC WD6400AAKS-0-1D05-596.17GB>
/scsi_vhci/disk@g50014ee2aedf73ce
4. c5t50014EE2AEDF7498d0 <ATA-WDC WD6400AAKS-0-1D05-596.17GB>
/scsi_vhci/disk@g50014ee2aedf7498
5. c5t50014EE2AF872299d0 <ATA-WDC WD6400AAKS-0-3B01-596.17GB>
/scsi_vhci/disk@g50014ee2af872299
6. c5t50014EE204411A53d0 <ATA-WDC WD6400AAKS-0-1D05-596.17GB>
/scsi_vhci/disk@g50014ee204411a53
7. c12t1d0 <ATA-WDC WD800JD-00HK-3G13-74.53GB>
/pci@0,0/pci15ad,7a0@15/pci1000,3040@0/iport@2/disk@p1,0

The first disk is the vmware disk OI is running on. The next 6 are the WWN names we already discussed. The new 80GB disk is 'c12t1d0'. That has me totally baffled as it IS on the same HBA!
 
Only thing I notice is the WD 640GB drives do have a WWN on the label, and WD 80GB drive does not. Still don't know why the controller number is different?
 
WWN must be assigned by Western Digital.
I suppose, your old disk does not have one
so LSI will display port id instead (There are only these two options)
 
Oh, that's interesting. Yes. the HBA seems to have assumed the 640GB drives are multipathed. Yes, thanks, I think I'll give this a try - it is more than a little confusing, and I have no interest in multipathing. Do you know why the controller number is different?
 
Hmmm, odd. I did 'stmsboot -d' and rebooted OI. Now, I see this (the new disk is not showing since I unplugged it):

0. c2t0d0 <VMware -Virtual disk -1.0 cyl 2085 alt 2 hd 255 sec 63>
/pci@0,0/pci15ad,1976@10/sd@0,0
1. c4t50014EE0AC01D8EDd0 <ATA-WDC WD6400AAKS-0-4C07-596.17GB>
/pci@0,0/pci15ad,7a0@15/pci1000,3040@0/iport@4/disk@w50014ee0ac01d8ed,0
2. c6t50014EE2AF872299d0 <ATA-WDC WD6400AAKS-0-3B01-596.17GB>
/pci@0,0/pci15ad,7a0@15/pci1000,3040@0/iport@8/disk@w50014ee2af872299,0
3. c7t50014EE2AEDF7498d0 <ATA-WDC WD6400AAKS-0-1D05-596.17GB>
/pci@0,0/pci15ad,7a0@15/pci1000,3040@0/iport@10/disk@w50014ee2aedf7498,0
4. c8t50014EE2AEDF73CEd0 <ATA-WDC WD6400AAKS-0-1D05-596.17GB>
/pci@0,0/pci15ad,7a0@15/pci1000,3040@0/iport@20/disk@w50014ee2aedf73ce,0
5. c9t50014EE0ABCEE0A9d0 <ATA-WDC WD6400AAKS-0-4C07-596.17GB>
/pci@0,0/pci15ad,7a0@15/pci1000,3040@0/iport@40/disk@w50014ee0abcee0a9,0
6. c10t50014EE204411A53d0 <ATA-WDC WD6400AAKS-0-1D05-596.17GB>
/pci@0,0/pci15ad,7a0@15/pci1000,3040@0/iport@80/disk@w50014ee204411a53

Still have funky WWN names, only the controller number is now different for each disk, and the device info now refers to pci instead of scsi_vhci.
 
You must log in or register to reply here.
Back
Top