[rescue] Sparc Classic & IPX rescue

Sandwich Maker adh at an.bradford.ma.us
Mon Dec 16 10:11:19 CST 2013

" From: Jerry Kemp <sun.mail.list47 at oryx.cc>
" Solaris 2.4 was where I got my start on Sun products, although I later worked 
" with 2.3 and SunOS 4.x/Solaris 1.x products.
" We deployed quite a few SparcStation 20's and SparcServer 20's running Solaris 
" 2.4.  After we got Solaris 2.4 installed, there was a follow on/included CD that 
" was full of patches.  For what we used those systems for, Solaris 2.4 was rock 
" solid and I don't remember ever having any system crashes.
" This was during my time in the Air Force, and these boxes were used for standard 
" Internet service(s), i.e. (DNS, FTP server, web server, NFS file shares, email, 
" RADIUS authentication for dialup modems) and Banyon Vines ENS (Enterprise 
" Network Server) directory service.
" These (Sun) boxes replaced older ATT 3b2 600G and 600GR systems.
" Good times.....

my first unix was sV at bell labs, on a variety of machines including
vaxes and an amdahl as well as att 3Bs.  sunos4 was my first sun
exposure, and i was at a heavy sunos4 shop when they decided to dip
their toes into solaris 2.  2.4 was current, probably a good thing.

that's also where i discovered jumpstart.  i moved the files to our
nas box and discovered one of the three boot config files -didn't-
escape <cr> with \ contrary to the manual...  i still recall the
thrill when i got all the questions pre-answered and the customization
scripts debugged and i could go from 'boot net install' to login

i was two weeks late getting the first batch up and rather the goat,
but quite the hero when i had the second batch up the day they
arrived.  all i had to do was enter their names and ips into the
jumpstart config, then walk around typing 'boot net install'.

i used the same trick on a y2k job later, got the js env tweaked, told
the group to take a long lunch - friday - then rolled their machines
from 4.1.4 to 2.6.  the thrill there was watching them come back, log
in, and go back to work.

also at the sunos4 shop, i worked out a way to semi-automate patching,
since i was iirc managing almost 100 machines and sunos4 patching was
supposed to be -by-hand-.  i had a script that fetched and parsed the
patchreport, then fetched patches i didn't already have, untarring
them in flight.  when run by a different name it removed old patch

then i created empty directory skeletons on the nas, for kernel
patches, openwin patches, and all the others, and i placed the
patched files in them.  the kernel tree could be rdisted and a new
kernel rebuilt any time though reboots had to be coordinated; the
openwin tree could be rdisted when i could get the users to log off
for a moment.  the rest had to be scheduled downtime, and the only
file i found i couldn't rdist was rdist itself, but if i did that by
hand the rdist job skipped over it.  then any time a new patch came in,
move it to the relevant tree and rdist again.  it tremendously reduced
patch time and effort, and individual machine downtime, and made me
feel very clever.

another fun thing was a kernel tuner script i wrote based on an old
sun performance overview article.  since sunos4 had paging, one of the
things it did was shut off the redundant and less efficient swapping.
cool to see swapping activity collapse while paging stepped -slightly-
when run on a busy system.  i think i still have it somewhere...
Andrew Hay                                  the genius nature
internet rambler                            is to see what all have seen
adh at an.bradford.ma.us                       and think what none thought

More information about the rescue mailing list