SGI: Hardware

D9 Stone FC-AL HDD mounting order ? - Page 2

Ok, looking through the responses, I guess I'm stumped.

The SES processor isn't likely assigning the FC-AL ID's; it's usually not that smart. The SES processor will give status information (and possibly some limited control) about the drives, but there's usually a hard configuration option in the shelf to give a starting address. For example, I seem to recall the older NetApp FC shelves or maybe the Clariion ones (it's been a long time) had either a dial switch or up/down switches to select the starting address. Then each drive bay is incremented.

Knowing that they show up differently in the hinv every time, then it sounds like whichever drive is spinning up first is assigned an FC-AL address first. I think there may have been an "auto" mode in FC-AL where if there isn't a drive position specified, one is assigned automatically, but those details are a bit fuzzy.

Does the DS9 have any switches on it anywhere? If so, that might configure the "auto" mode. And also, has the DS9 always misbehave this way, or did it just recently start misbehaving?

Chris
:O2000R: (<-EMXI/IO6G) :O200: :O200: :O200: (<- quad R12k O200 w/GIGAchannel and ESI+Tex) plus a bunch of assorted standalone workstations...
The Keeper wrote: Knowing that they show up differently in the hinv every time, then it sounds like whichever drive is spinning up first is assigned an FC-AL address first. I think there may have been an "auto" mode in FC-AL where if there isn't a drive position specified, one is assigned automatically, but those details are a bit fuzzy.


They don't show up differently in the hinv, they appear identical every time :

Code: Select all

Integral SCSI controller 0: Version QL1040B (rev. 2), single ended
Disk drive: unit 1 on SCSI controller 0 (unit 1)
Integral SCSI controller 1: Version QL1040B (rev. 2), single ended
Integral SCSI controller 4: Version Fibre Channel AIC-1160, revision 4
Disk drive: unit 0 on SCSI controller 4 (unit 0)
Disk drive: unit 1 on SCSI controller 4 (unit 1)
Disk drive: unit 2 on SCSI controller 4 (unit 2)
Disk drive: unit 3 on SCSI controller 4 (unit 3)
Disk drive: unit 4 on SCSI controller 4 (unit 4)
Disk drive: unit 5 on SCSI controller 4 (unit 5)
Disk drive: unit 6 on SCSI controller 4 (unit 6)
Disk drive: unit 7 on SCSI controller 4 (unit 7)
Integral SCSI controller 3: Version Fibre Channel AIC-1160, revision 4


The disks ids are always sequential starting from 0 but they are not assigned to the same physical disk in the enclosure, which seems totally strange and illogical to me.

For example, when I list the content, what was previously mounted on /stor0 at last power-up is now mounted on /stor1 and conversely (I mounted the 8 disks on /stor0 - /stor7 folders). Just as if I swapped the disks in the enclosure.

The Keeper wrote: Does the DS9 have any switches on it anywhere? If so, that might configure the "auto" mode. And also, has the DS9 always misbehave this way, or did it just recently start misbehaving?

Chris


There's no visible switches or jumpers anywhere on the D9 enclosure. I'll have to dismantle it completely and check the innards. Even if I find random jumpers inside, I'd be wary of changing anything beause I have no documentation about the D9 Stone hardware and couldn't find any on the net. I guess it always behaved that way but I just didn't see it before because at first, I didn't use all the disks and also, I hadn't set up a website with symbolic links pointing to data on several disks. Thanks for your help in any case ! 8-)
:Octane2: :Octane: :Octane: :O2:
The issue you are describing like sounds you need to configure a persistent binding so that a particular WWN binds to the same SCSI id each time. I've done this in the in past on Solaris machines through the driver config file but not sure where that is for IRIX.
Sorry for the delayed response. WWN bindings under IRIX are typically only when you're operating in a fabric with a switch. When the drives are direct-attached, the drives look more like SCSI.

Now that I see your hinv (or maybe I didn't notice before), I see you're using the Adaptec AIC-1160. I've never used that one, I've only ever used QLogic cards. Maybe the AIC-1160 is being stupid, I'm not sure. The LUN mapping should be persistent across the FC-AL ID's, so there might be a firmware configuration setting, but it'd be hard to get to in an IRIX box. Adaptec got out of the Fibre Channel and iSCSI business a very long time ago, possibly because they couldn't figure out what they were doing.

Does the AIC-1160 have a DB9 connector or an HSSDC connector? Either way, I'd probably suggest trying a QLA2200, because I never ran into the problem you're describing. I could tell from a Fibre Channel analyzer trace, but, well, those aren't very common.

Tell you what, drop me an email at my gmail account, with the username kalisiak. I'm sure I have copper QLA2200's laying around, and I can take a look and see if I have an extra HSSDC-DB9 cable if you don't have one.

Chris
:O2000R: (<-EMXI/IO6G) :O200: :O200: :O200: (<- quad R12k O200 w/GIGAchannel and ESI+Tex) plus a bunch of assorted standalone workstations...
The Keeper wrote: Sorry for the delayed response. WWN bindings under IRIX are typically only when you're operating in a fabric with a switch. When the drives are direct-attached, the drives look more like SCSI.


Don't be sorry, please ! I'm certainly not in a hurry and any kind of help is warmly welcome ! 8-)

The Keeper wrote: Now that I see your hinv (or maybe I didn't notice before), I see you're using the Adaptec AIC-1160. I've never used that one, I've only ever used QLogic cards. Maybe the AIC-1160 is being stupid, I'm not sure. The LUN mapping should be persistent across the FC-AL ID's, so there might be a firmware configuration setting, but it'd be hard to get to in an IRIX box. Adaptec got out of the Fibre Channel and iSCSI business a very long time ago, possibly because they couldn't figure out what they were doing.

Does the AIC-1160 have a DB9 connector or an HSSDC connector? Either way, I'd probably suggest trying a QLA2200, because I never ran into the problem you're describing. I could tell from a Fibre Channel analyzer trace, but, well, those aren't very common.

Tell you what, drop me an email at my gmail account, with the username kalisiak. I'm sure I have copper QLA2200's laying around, and I can take a look and see if I have an extra HSSDC-DB9 cable if you don't have one.

Chris


The AIC-1160 was there as part of the original Avid config I think. It has a DB9 connector and the D9 Stone also has a DB9 connector. Thanks a lot for your proposal, you'll be receiving an email any minute now ! :lol:
:Octane2: :Octane: :Octane: :O2:
The AIC-1160 is not SAN capable if I'm not mistaken, but for a direct attached (FC-AL) JBOD it should do just fine. I used to have one in my O2K deskside with a monster SUN A5200 array attached.
Now this is a deep dark secret, so everybody keep it quiet :)
It turns out that when reset, the WD33C93 defaults to a SCSI ID of 0, and it was simpler to leave it that way... -- Dave Olson, in comp.sys.sgi

Currently in commercial service: Image :Onyx2: (2x) :O3x02L:
In the museum : almost every MIPS/IRIX system.
Wanted : GM1 board for Professional Series GT graphics (030-0076-003, 030-0076-004)
I guess at this point I can't remember. Is the SGI AIC-1160 a PCI card or an XIO card? If it's an XIO card, then yeah, it would be a little challenging to swap for a PCI card... :-/
:O2000R: (<-EMXI/IO6G) :O200: :O200: :O200: (<- quad R12k O200 w/GIGAchannel and ESI+Tex) plus a bunch of assorted standalone workstations...
jan-jaap wrote: The AIC-1160 is not SAN capable if I'm not mistaken, but for a direct attached (FC-AL) JBOD it should do just fine. I used to have one in my O2K deskside with a monster SUN A5200 array attached.


It is directly attached in my case. Still, it's working weirdly IMHO. Maybe it's a firmware issue. I have another FC-AL XIO card to try but it is an Origin/Onyx2 kind of hook so I need to hack it first before trying.
:Octane2: :Octane: :Octane: :O2: