From: Dominique Petitpierre Subject: Mac won't boot from second disk. Why? Here is the summary of the answers I got for the question I asked on December 18th in Info-Mac Volume 13 Issue 176: |I have just added a second internal disk drive in my Macintosh, and would |like to make it the boot disk while keeping the old disk still bootable in |case of problems. Infortunately, whatever I do, the Mac insists to boot |from the old disk. |- Why? What can I do? The short answer: ================= What worked for my purpose was to exchange the SCSI ids of the disks, putting the new disk at target 0, and the old target 0 disk at target 1. Now I can use Startup Disk to choose whichever of these two disks as boot disk, and at the next restart the mac boots from the chosen disk. The reason it works has to do with the fact that the Mac tries longer to boot from SCSI target 0. Many thanks to Julian Daniel, North Star Computer Ltd. who suggested this. If my purpose is fullfilled, the real problem is not solved: How can you force the Mac to wait long enough for disks not at SCSI target 0. For example I have another old external disk, that I use only briefly for backups because it is too noisy. The same symptom occurs: even if it is bootable when it is the only disk on the SCSI bus, the Mac won't boot from it if the internal disk is available, whatever I do. Thanks to all the people who helped with suggestions: Julian Daniel Jim Carr Christian F. Buser Gregory Johnson Glenn Schaffer Thomas Ytterberg Dr. Francisco-J. Renero-C. Art Treiber The long answer: ================ For the curious, here is a summary of other suggestions I received with comments and of a few things I discovered while trying to solve this. First here are the informations I provided with my original question: |Here are some facts and things I tried: | |My setup: |Computer: MacIIx with System 7.0.1 and Tune Up 1.1.1 (on the old disk). |Old disk is a Quantum 170S at SCSI target 0 with driver FWB HDT (I never |touched this, and would prefer not to). |New disk is an IBM DPES 31080 at SCSI target 3 with Silverlining 5.4.2/16 |(also tried 5.6.3/25) |The new disk has a bare fresh system 7.0.1, installed from the Apple System |7.0.1 diskettes. |The "Startup Disk" control panel shows the new disk selected as startup disk. | |Tests and observations: |- The System Folder on the new disk is blessed (shows the icon). |- After boot on the old disk, the new disk mounts fine (appears on | the desktop). |- If the old disk is disabled (physically disconnected), the new disk boots | fine (so it IS bootable). |- At boot time, I hear that the new disk is probed twice: once before boot | starts on the old disk and once after. With Silverlining 5.6.3/25 this is | confirmed by the caption that Silverlining displays twice on the | screen. It seems also that in the first round, the new disk (SCSI 3) | is probed before the new disk (SCSI 0), which is what I expect. | |Things I suspect: |- Answer from the second disk is too slow for the impatient MacIIx | (But shouldn't disk of recent make be faster than old ones?) |- Incompatibility between disk drivers (FWB and Silverlining). - The most common suggestion was to check SCSI termination. In fact, during my trials, I had once a clear case of bad termination due to a bad connection: the old disk would eventually boot after waiting ages at each stage of the boot, but the new disk would not even mount. MtEverything reported bad termination. So my problem might have been a lighter case of bad termination: timeout for boot, but not preventing mount. I was very careful with that when I installed the new IBM 1080 disk: I made it the last device on the chain inside the Mac and enabled its active termination (jumper). On the Quantum 170S, I took care to remove the three resistor packs next to the SCSI connector to disable termination. I was pretty sure that there was nothing wrong with termination because I was able to connect and use without other problems seven devices on the SCSI chain (two internal disks, two external disks, a tape drive, a CD ROM drive and a scanner), I doubled checked it by shortening the chain to just the two internal disks and terminating the open ended side of the chain with a pass through terminator on the SCSI plug of the Mac's back panel. It made no difference. My next move would have been to measure the termination resistance with a multimeter (see how to below). - The second most common suggestion was to check the batteries: Mac II, IIx and IIfx have two internal lithium batteries, for power up and for keeping the PRAM settings and the time. When I checked with a multimeter, both batteries provided the expected 3.6 volts. Because of the old age of my MacIIx, I decided to buy a spare pair of new batteries, to have handy when things go bad, and tried them as well: it made no difference. Good thing I saved my PRAM settings during another test (see below), because I had overlooked the fact that I would loose them when taking off the battery. - Another suggestion was to zap the PRAM (not forgetting to save the settings before with a tool like Tech Tool 1.0.9). It did not make any difference. - Some suggested to make the old disk not bootable either by moving the System and Finder in separate folders, or by specifying with Silverlining or HDT that it should not be mounted at startup. Not very practical; that would have been my last resort: In the second case you need a mounting utility to force mount after startup. And in case of trouble with the new disk, a floppy is necessary to make the old disk bootable again. - Some people suggested to use HDT on the second disk as well: it did not work (with an old HDT version 1.0). Worse it put my Mac in a very strange state where it froze in the middle of the boot with a blank screen. The only way to force it to boot from an emergency floppy was to use the magic key combination command-option-shift-delete, you know, where you have one hand on command-option-shift on the keyboard, the other hand behind the computer case to press the restart button, and the nose on the delete key :-)... Worse, my first level safety net failed: Silverlining on the emergency floppy would cause a bomb immediately (I checked it afterwards it was just fine, so HDT must have really messed things up...). I had to resort to a system 6.0.8 floppy disk and another Silverlining floppy and play disk jockey to alternate between the two floppies. Moralite': have alternate rescue solutions.... - I was too chicken to replace the HDT driver on the first disk with Silverlining. (Yes I have backups... 150 MB worth of compressed archives on floppies and tapes, not a nice perspective to reinstall :-(.) Once I move the old disk data on the new one, and make a new backup, I might dare change the driver on the old disk. - Jim Carr suggested: >HDT Utils (Part of Hard Disk Toolkit) has an option >to set startup delay on internal drive. This would allow extra time for >your prefered drive to spinup. Not an option for Siverlining controlled drive, and I was not really keen on retrying for a third time the HDT driver (see previous point). Today I bought HDT 1.7 Personal Edition, and will see if I can do something for the slow external disks. (I am fed up of taking apart the internal ones to fiddle with jumpers :-), now they work as I want and I won't touch them for a while). - Glenn Schaffer suggested: >Try to swap the order the drives are plugged in to the motherboard. Haven't tried it, because my SCSI ribbon does not let me to. - and also >spin-up time needs to be accommodated. >I recall a few applications which can do that (saved >in PRAM, if I remember right), but don't have any of them or recall their >exact names. Something like "Startup Delay" ??? I don't think the >startup delay is your problem unless you've got a lot of cards in you Mac >which pull a lot of power (the IIx has a pretty hefty power supply)." The only tool with a similar name I found was "Delayed Startup Items" which addresses a different problem. According to Jim Carr, HDT Utils should allow this. - I also double checked the jumper settings on the disks; In particular on the IBM DPES 31080, jumper 7 disables "target initiated synchronous negotiation". It did not change anything, on or off. I did not dare to put on jumper 4 to disable auto spin. Would it bootor mount if autospin is disabled? Diagnostic tools and informations ================================= various things I found useful (or could have) in my search for a solution (program file pathnames are given for Info-Mac archives mirrors). - Mt.Everything 1.0.3, disk/mt-everything-103-cdev.hqx Complains explicitely in case of bad termination. - SCSI Probe 3.5, disk/scsi-probe-35.hqx Suggested by Glenn Schaffer for detecting bad termination. Cannot selectively mount disks like Mt.Everything-103. - Tech Tool 1.0.9, cfg/tech-tool-109.hqx Suggested by Jim Carr to save and zap the PRAM. - TattleTech 2.2.0, cfg/tattle-tech-22.hqx Useful to check the existence of boot blocks and system files on disks. Provides a nice concise description for each feature it reports. Can also give you the list of all your applications with their version and creator type. On the new disk I want to do a complete re-installation of my system and applications from original distributions, and such a list will enable me to check that I get the right versions and that I apply all the upgrades! - Set Startup 1.3, cfg/set-startup-13.hqx A replacement for Startup Disk that let you specify the startup disk by SCSI id, even if the corresponding disk is not mounted. I haven't used it. - System 6.0.8 and System 7.0.1 (ftp://ftptoo.support.apple.com/pub/apple_sw_updates/US/mac/system_sw/ in system_6_0_8_1.4m and system_7.0.1.) 1.4 MB disk images of Systems, practical if, like me you are not sure to be able to read 800k floppies anymore. Also System 6.0.8 is small enough to let you put easily other things on the floppy: Silverlining, SCSIProbe etc. - Larry Pina's book "Macintosh II Repair and Upgrade Secrets", published by Hayden Books, helped me for the battery checking. - The Web page of Quantum Corp., where I found documentation on disk jumper settings (http://www.quantum.com/support/jmprmenu.htm). Infortunately my Quantum 170S was not described there, but I could work from analogy with for example the "Empire 540", which shares some jumper two letter labels. - The Web page of LaCie, where I found an updater for older Silverlining software (http://www.lacie.com/updates.html), information on SCSI termination (http://www.lacie.com/jumpers.html), a very well made "Encyclopedia of Hard Drive Terms and Concepts" (http://www.lacie.com/data.html) explaining SCSI issues and hard drive formatting, a summary of keyboard command tricks (http://www.lacie.com/tip.html), and a page on troubleshooting (http://www.lacie.com/shoot.html). - The SCSI Frequently Asked Questions for the comp.periphs.scsi newsgroup, (ftp://rtfm.mit.edu/pub/usenet-by-group/comp.periphs.scsi/*) The following message I collected long time ago about checking passive termination with a multi-meter: >Date: Wed, 21 Jul 93 15:03:24 GMT >From: wyman@esfm01.SINet.SLB.COM >Subject: (R) SCSI device termination - how to check for it with a multi-meter > >..... > >2. Even without connecting the device and without opening it, a simple test >with a cheap multi-meter (or an expensive one, if that's all you have!) >will reveal whether your device is internally terminated: > >With the device power off, measure between pins 26 & 50 of the 50-contact >SCSI connector, "the narrow side" of the D-shaped ribbon-contact connector. >The measured value for an un-terminated SCSI device will be 10,000 ohms >or more, effectively an open circuit. > >A device with standard passive resistor packs installed will measure >about 260 ohms plus or minus 15 ohms. A doubly-terminated device >(internal resistor packs + external terminator on the other connector) >will measure about 130 ohms (half of the normal value). Of course, in >this latter case you MUST remove one of the terminators to be correctly >configured. I make a practise of checking ALL in-coming SCSI hardware >and removing the internal SCSI packs if found. Other posters have >exhaustively explained the correct configuration for a SCSI bus. I won't >repeat that discussion here. > > > 25 1 > ________________________________________________________ > ( o o o o o o o o o o o o o o o o o o o o o o o o o ) > \ --------------------------------------------------- / > \ -------------------------------------------------- / > \ o o o o o o o o o o o o o o o o o o o o o o o o o / > -------------------------------------------------- > 50 26 > > ^ ^ > |__________________ 260 ohms ___________________| > >Note that you can use the procedure to examine a chain of SCSI devices >for correct termination: Disconnect the System SCSI cable at the first >device and make the measurement. You should only "see" one terminator >in the chain, at the far end. If you then remove it from the last >device, you should see an open or nearly-open circuit as above. > >I use the same procedure for checking SCSI device chains on SUN and >VAX workstations. > >Doug.Wyman >Schlumberger - GeoQuest >Montrouge, France Conclusion: =========== I hope these informations will be useful to people that have boot problems. And in case you were wondering, I paid $290 for the brand new IBM 1080 MB disk, without cable etc.; worth it, even on an aging Mac IIx :-) Dominique -- Mr. Dominique Petitpierre | petitp@divsun.unige.ch | Tel: +41 (22) 705 7117 ISSCO, University of Geneva/54 route des Acacias/CH-1227 GENEVA (Switzerland)