home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!sun-barr!news2me.EBay.Sun.COM!exodus.Eng.Sun.COM!appserv.Eng.Sun.COM!appserv!limes
- From: limes@straylight.eng.sun.com (Greg Limes)
- Newsgroups: comp.sys.sun.admin
- Subject: Re: 630MP boot problem revisited
- Date: 20 Jan 93 22:02:10
- Organization: Sun Microsystems Computer Corporation (Mountain View, California,
- USA)
- Lines: 40
- Distribution: world
- Message-ID: <LIMES.93Jan20220210@straylight.eng.sun.com>
- References: <1993Jan17.145822.15288@coral.uucp> <llrajpINNdcq@appserv.Eng.Sun.COM>
- NNTP-Posting-Host: straylight
- In-reply-to: weinberg@sho.Eng.Sun.COM's message of 20 Jan 1993 19:39:37 GMT
-
- In article <llrajpINNdcq@appserv.Eng.Sun.COM> weinberg@sho.Eng.Sun.COM (Glenn Weinberg) writes:
-
- | In article 15288@coral.uucp, jdc@coral.uucp (James D. Corder) writes:
- | > SCSI ID 3 is sd0. Sun stated that they once had a
- | >good reason to do this. However, I have yet to meet a rep who knew what that
- | >resson was, and why they still do it:-)
- |
- | So you can explain it to your rep the next time you see them :-)
- |
- | When the SPARCstation 1 came out (almost 4 years ago now!), there were lots
- | of old "shoebox" SCSI expansion boxes out there. Each shoebox could contain
- | two devices. Since older "desktop" systems such as the 4/110 had no
- | provision for internal disk drives and this was in the days before embedded
- | controllers, the first shoebox typically used target 0, luns 0 and 1, and the
- | second used target 2, luns 0 and 1. It was decided to make life "easier" on
- | SPARCstation 1 customers who wanted to continue to use their existing
- | shoeboxes by making the internal drives be targets 3 and 1 to avoid
- | conflicting with the existing target usage.
-
- And, to make it worse, the 4/330 -- which came out at the same time -- solved
- the same problem in almost but not quite the same way ;-)
-
- Same solution on the target IDs, we wanted old shoeboxes at ID=0 and
- ID=2 to still work as expected.
-
- However, different solution on what UNIX called them. The SS1
- scrambled the names of the disks, so that targets {3,1,2,0} were
- {sd0,sd1,sd2,sd3}, while the 4/330 kept the old Sun-4 mappings of
- targets to names, {0,1,2,3}=>{sd0,sd2,sd4,sd6}.
-
- So, on the SS1 (and cousins), you boot from sd0 just as you would
- expect, but the target is #3. Frankly, I have no idea what an SS1
- would do if fed a double-disk shoebox (ie. target #0, lun #1).
-
- And, on the 4/330 (and cousins), your old shoebox at ID#0 is sd0, and
- your other shoebox at ID#2 is sd4, just like always, but now you boot
- from sd6 at ID #3 ...
- --
- Greg Limes [limes@eng.sun.com]
- Not speaking for Sun Microsystems Computer Corp., Mountain View, California
-