init's untimely death.

[Gary Blumenstein]

In article <2045 at egvideo.UUCP> edhew at egvideo.UUCP (Ed Hew) writes:
>In article <1989Jun21.114506.1378 at tapa.uucp> larry at tapa.uucp (Larry Pajakowski) writes:
>> Perhaps someone can shed some light on a perplexing problem we are having.

Me three!  I had a similar experience with init dying.  Listen to this one.

Some months back I hooked up a serial line from one of our ports to a port
on our VAX so I could log into our VMS (ugh!) system.  Well one night I was
showing the Operator in the data center what Usenet was all about and I
said, "wait a minute, I'll give you an account on our XENIX system so you
can log in when you're bored silly at 11pm and read a little comp.os.vms
or whatever".  "Really!?", he said.  "No problem, we have a direct line
set up ready to go.", I said as I dim-wittedly enabled the port on the
XENIX side.

Enter the dreaded battle of the logins.  In this case it was VMS LOGIN versus
XENIX getty.  Invisibly, the poor computers were duking it out "behind the
scenes" with each program interpreting each other's login message as INVALID
login id's.  In this case XENIX was the looser with init getting trashed
every so often.  What made matters so frustrating for me was A)  I had
completley forgotten about enabling that darn VAX line and B) init
would die inconsitenly at random intervals.  Sometimes the system wouldn't
go down all day, at other times I'd be rebooting 3 or 4 times per day.

The sysptoms were all classic, just as the others had described.  init would 
die leaving every terminated process defunct and without a parent and getty's 
could not spawn after a user exited their shell.

Not realizing the stupid mistake I made by enabling that line, I was at a 
real loss trying to figure out what went wrong.  I had become convinced that
somehow the kernel had been corrupted and I had just began the process of
reinstalling the link kit and drivers.  As I was disabling the serial ports so
I could boot off a backup kernel, thats when I noticed the enabled line.
Aparrently, the Operator never had an opportunity to try the line anyway.

I'm almost embarrased to tell that one.  Have mercy!

- gb








-- 
Gary Blumenstein, UNIX Systems Administrator // CIBA-GEIGY CORPORATION, USA
===========================================================================
Voice: (914) 347-4700                  7 Skyline Drive, Hawthorne, NY 10502
FAX  : (914) 347-5687     uucp: ...{philabs, gaboon}!crpmks!{sysadm, garyb}