Boydie wrote:... the first thing you should do with a new machine was make a complete set if backup floppies ... If you're very lucky, this will be what your backup floppies are and there won't be any personal data.
'Fraid not - the dump is dated and it's too late to be a factory-fresh R140 and also too late to be a "just done the upgrade to 1.15" dump.
What a fun morning it's been: Syquest died, dug out the spare* and set that up (less convenient as it's ID needs to be jumper-set rather than using a button on the back of the box), started with a fresh cartridge (if you can call something stamped "OCT 1996" fresh), sectioned/partitioned, booted 1.21, updated the disktab file** and ran newfs, ran my bodged mknewfs2 script, hand-copied vmunix, dsplit and libC rather than rely on cpsys, trimmed cpsys so as just to do the archive extraction.
Time to reboot, now into 1.15 - but it gave a kernel panic first time and now freezes, with the Syquest activity light on constant orange, on subsequent tries with 1.15 (device sd2) or 1.21 (device sd0). Ctrl-reset to get out of that. I can run the filesystem check on sd2a from the !RISCiX desktop app, for what that's worth, and it comes out clean.
Last try will be with the internal drives unpowered and the SCSI card terminated, for which I'll have to re-jumper the Syquest to ID0.
I don't understand why everything seems fine while setting up the installation, yet both it and then 1.21 refuse to boot.
UPDATE: Most probably a cable / termination / other SCSI voodoo issue - with Syquest as the only device on the chain, and a terminator block instead of a drive's "internal" termination, 1.15 has just booted fine - and I did not get the "cannot reboot with this rom" error I saw yesterday.
Not a "pure" 1.15 set-up, as I'm using RISCiXFS 1.24 and RO3.11. Aiming to try the restore stage this evening.
UPDATE2: restore from 19 floppies worked, so I now have the system as it stood in August 1993! The manual says a full backup can take 45 floppies and 5 hours; restoring 19 took 35-40 minutes, which I guess reflects ARM3/SCSI vs ARM2/ST506. Unfortunately no packages installed under packageadmin, and the first one I tried (X11R2) failed with a CRC error. And various hard-coded references to st0a don't help now it's running on sd0a. BUT this setup looks small enough (15.8MB per df) to fit on a 20MB ST506 ....
* There is a difference between the dead Syquest and the spare is a firmware revision - spare is 3.05 rather than 2.04.
** newfs returns an EOF error using a disktab that defines pa (partition 0's size), but the dynamic approach (as used for RO3065) works, and gives a clean fsck. I therefore suppose any drive can be added using only C/H/S, sector size and block/file parameters - meaning you don't need to update disktab if you alter the partition sizes on that drive, or want to add another drive with same geometry but different partitioning.