BOOT SHOCKER! JAZZMAN KO'S ELI IN MULTIUSER UPSET!

gistfile1.txt

  BOOT SHOCKER!  JAZZMAN KO'S ELI IN MULTIUSER UPSET!

  In a sold-out Lab 3219, league veteran and 20 MHz war-horse, jazzman, defeated
  the 40 MHz favorite, eli, in a PROM-to-multiuser sprint.  jazzman's time of
  four minutes, thirty-eight seconds soundly defeated eli's lethargic five
  minutes, fifteen seconds.

  While the match was at times close, jazzman only stumbled early.  "jazzman
  was slow bringing up interfaces; I should have caught him then, but I
  couldn't seem to fire on all cylinders," reflected an exhausted eli.
  Assessing the sluggish performance, eli noted, "having my cache disabled
  early in boot definitely hurt me out there today", adding "by the time I
  regrouped and got the cache turned back on, this contest was over."

  "If nothing else, today shows that cache still matters in this league," said
  a clearly elated jazzman, adding "eli showed a lot of character today; he
  booted with tremendous heart."

  jazzman, fresh off victories over perennial heavyweights wopr-2, ugly,
  jurassic, and now eli, faces a tough schedule in coming weeks.  After
  a conference showdown with voodoo-chile, jazzman will face acid in an
  exhibition match before squaring off against league powerhouse, chugwater.

  Despite the upset over eli, chugwater remained indifferent.  When asked
  about the upcoming bout with jazzman, chugwater replied, "hasn't he been
  EOL'd?"

  jazzman dismissed rumors that he had been EOL'd as "obviously false", adding
  "you don't EOL a platform booting at the top of its game."

This was written on April 27, 1999; the cyclics wad had sun4c-specific components, and I was (apparently) bored waiting for sun4c test machines to boot. (These were all machines in Lab 3219 in MPK17 in Menlo Park, a space now occupied by Facebook.) Like some neolithic garbage dump, there is tons of history embedded in this: eli was an ancient Viking microprocessor that had an infamous defect that required its instruction cache to be disabled during boot; wopr-2 was our E10K that took absolutely forever to boot in 1999; acid was one of the Galaxy (SPARCserver 630MP) machines beloved by the late, great Roger Faulkner; ugly was an x86 machine whose name insulted its appearance (and included brethren heap, pile, lump, and plop -- all named by Roger); jazzman and voodoo-chile both long predate me and I believe that they may have been named by Dock Williams; and chugwater was (if memory serves) an Ultra 60 that was the state of the art at the time. Sadly, I'm not sure that server boot times have improved much in the last two decades; plus ça change!

I remember wopr-2. We had downstairs arachnid (8-way sun4d... 8-way, 8-legs, spider, y'know...). And yeah, the boot times always sucked.

One hates to disagree with Roger Faulkner, but before I knew you (and taken out of service before we got seriously involved in Platinum Beta) we had a Galaxy, a 4/690MP (doctor-seuss) with all the trinkets including, at one point, some weird NFS co-processor (a rebranded Interphase board). We'd be sold it for about ten dollars as a spoiler when we were threatening to buy our first Auspex. It was a complete and utter dog running SunOS, and far, far worse when we had something like 2.4 on it. The hardware was admirable and, I gather, it was the team that had gone off to design the Auspexes as they were physically very, very similar. But the overall package was shamefully bad, particularly as compared to its predecessor 4/470 (kether) which was a delight.

About a year later we bought an Auspex anyway. And then Bonwick produced 2.5.1, which probably would have solved our problems.