to: customerservice@a_used_Mac_vendor subject: problems with my mac Hi, I bought a used Mac 7300 from you back in February (order omitted) and it's been working pretty well for some weeks, running OS X. It took some debugging though, and I thought I'd mention the problems I had, in the hopes that you might be able to avoid them in future. Customers with less experience debugging hardware and software problems than I would likely be baffled by what I encountered. (I was going to recommend that my sister, not a computer person at all, buy a Mac from you, but I am most reluctant to do so after my experience, particularly since she lives on the other coast, so I can't just fix it for her.) 1. The main (internal) SCSI bus wasn't terminated. OS 9 didn't *seem* to care but OS X behaved strangely until I terminated the bus using the disk's jumpers. As I recall, both OSs were prone to mysterious hangs until I terminated the bus. 2. The disk, a Seagate ST15230N is "unsupported" according to OS 9 Disk Setup and the OS 10.2.0 kernel, but is supported by the 10.2.3 kernel. This made getting OS X installed tricky. I ended up copying the Seagate's contents to a Quantum XP32150, which is supported, and made that my main disk. I also added a 9GB IBM disk, since the 2GB Quantum was (and is) tight for a root disk. 3. The file system on the original disk was HFS, which OS X can't use. Both OS 9 and OS X can use HFS+ file systems, so HFS+ would be a more useful file system type to provide. I eventually bought the FWB Toolkit to solve various disk-related problems, and it turned out to be invaluable. My notes say ``apparently I need HFS & HFS+ partitions on a drive I want to install OS X on: OFW seems to only be able to boot from HFS [this could have been a dead-PRAM-battery problem, though] but xpostfacto at least wants to install into HFS+. The OS X installer can't cope with 640x480; i need a better monitor to do it.'' 4. Although you had added the Sonnet PPCG3-500 card (thanks!), you hadn't removed the L2 cache card and hadn't installed the Sonnet software from their floppy. 5. The OpenFirmware variables were set strangely (input and output were set to use a serial port instead of the screen and keyboard), but this was probably due to the battery that keeps the NVRAM (PRAM) alive having died (not surprising since the machine was manufactured 7 years ago and apparently still had its original battery, stamped "97"). I replaced it for about $6 with a new one from Other World Computing, and that cured all sorts of weird problems I had been having, including: - attempts to reboot the machine would usually make it shutdown the operating system, chime briefly and then just sit there, requiring a manual reset or power-cycle; - after a reboot, the NVRAM/PRAM contents often weren't what they had been set to, just moments before; - PCI cards behaved strangely and often weren't usable. Making sure that the PRAM battery is live should save lots of customer headaches. 6. A Mac that old needs some obscure adapters to connect to current hardware: - an RGB/VGA adapter for any non-Apple monitor - an Apple-SCSI/real-SCSI cable for any non-Apple external SCSI devices - an ADB/DB9 serial cable for any serial port (especially when PRAM variables are set wrong and you need to hook up a serial console). These are all a little hard to track down but are pretty inexpensive; providing them, perhaps at extra cost, should generate customer good-will. 7. The version of OpenFirmware is pretty old (1.something). It would be helpful to provide a little cheat-sheet of common OpenFirmware commands, since it's not easy to find the right documentation for *current* OpenFirmware, let alone 1.something OpenFirmware. I enclose my own cheat-sheet; feel free to use it (or not) however you wish. Geoff Collyer --- OFW cheat-sheet --- references: www.openfirmware.org bananajr6000.apple.com/ playground.sun.com/1275/ www.netbsd.org/Ports/macppc/ www.macdevcenter.com/pub/a/mac/2003/02/18/secure_tibook.html developer.apple.com/technotes/tn/tn1061.html www.netneurotic.de/mac/openfirmware.html diveintoosx.org/OpenFirmware.html typed during the boot chime: command-option-O-F enters openfirmware may need to run System Disk first to set up nvram variables for input (kbd) and output (/chaos/control; screen) command-option-P-R during 3 boot chimes zaps PRAM (or control-opt-p-r, but I think cmd-opt-p-r worked for me). command-option-n-v allows editing of nvram command-shift-option-delete supposed to boot off cd command-v shows boot messages; also "boot -v" command-s goes into single-user mode any time: control-option-power emergency off type at the ofw prompt (`ok'): devalias cd /bandit/gc/mesh/sd@3 setenv variable value nvsetenv variable value reset-all mandatory; really, really stores variables in NVRAM # partition numbers seem to start at 1 # beware need for "blessed" boot directories boot scsi-int/@0:4 # boot from disk at target 0, partition 4 boot scsi-int/@0:9 # "", partition 9 (more common) boot scsi-int/@1:6 # 2gb hfs+ boot scsi-int/@1,0:6,\mach_kernel # maybe boot scsi-int/sd@1:6 # is also supposed to work boot scsi-int/@3,0:9 # attempted boot off cd on scsi target 3 show-devs [node] load-base; dload file # load file via ether go not working correctly on the 7300 when I last tried them (may be better with working PRAM battery): "c" at reboot should boot from cd-rom - it pokes around on the cd, but doesn't boot from it correctly option at reboot should yield choice of startup disk not in our ancient OpenFirmware: probe-scsi [node] sift-devs sub-string # greps device tree for sub-string from Xpostfacto, when booting OS X installer: boot-device: scsi-int/@0:10 boot-file: -i boot-command: 0 bootr -v rd=*scsi-int/@3:9 or maybe boot-command: 0 bootr -v rd=*scsi-int/@3:9 boot ---