Cold Starts

Thread View: alt.folklore.computers

2 messages

[T] Back to Threads [V] Tree View [G] Group Articles

2 total messages Started by mcg... Darin McGrew Wed, 13 Dec 1989 20:42

#177

Author: mcg... Darin McGrew
Date: Wed, 13 Dec 1989 20:42

34 lines
2287 bytes

I had a summer job operating the EMCS (Energy Monitoring and
Control System) at Fort Ord (CA).  It consisted of two HP 1000's
connected by a proprietary interface, and hundreds of Z80 based
field units.  Typically, one HP 1000 did the processing and the
other handled communication with the field, although either could
be run by itself and handle both jobs.  The field units had a
battery backup that lasted roughly 30 minutes, after which they
had to be reinitialized by the master HP 1000(s).

This worked great, unless the entire base lost power for more
than half an hour, and this happened at least once or twice a
summer (typically a garbage truck driver would leave the forks up
and snag a power line).  All the field units would exhaust their
backup power, and need to be reinitialized.

Once power was restored, the reboot sequence would start
initializing the field units.  While the field units were running
some sort of self diagnosis, the master system would start
initializing more field units.  Later, when the field units
started reporting back to the master system, the HP 1000 handling
communication would get overloaded.  The other HP 1000 would wait
for its partner to acknowledge a message, then timeout and reboot
the system.

The only solution was to turn off scanning on all the field
units, but by the time the terminal I/O routines were started
there was only time to turn off scanning for about half a dozen
field units before the imminent reboot.  It usually took two of
us an entire day to disable scanning on all the field units, and
then it took me a couple days to restart scanning one field unit
at a time.

Darin McGrew			mcg...@Eng.Sun.COM
Affiliation stated for identification purposes only.

Re: Cold Starts

#178

Author: col... Sean Colbath
Date: Fri, 15 Dec 1989 22:45

36 lines
2816 bytes

In article <129...@sun.Eng.Sun.COM> mcgrew@sun.UUCP (Darin McGrew) writes:
>I had a summer job operating the EMCS (Energy Monitoring and
>Control System) at Fort Ord (CA).  It consisted of two HP 1000's
>connected by a proprietary interface, and hundreds of Z80 based
>field units.
<deleted story about the problems involved in sanely bringing up a
 distributed system>

Although not very similar, your subject line reminded me of an interesting
incident that occured at an academic site that I was involved with.  This
site was a large IBM VM shop, with most of the computing done at the central
campus.  However, they had several smaller machines at the other campuses
that ran unattended.  The operator at the main campus ran the machine
remotely.  Of course, the only work that couldn't be done remotely was
printer feeding and the rare tape mount, so students were hired to tend to
these necessities.  However, they were not expected (read allowed) to do the
normal operator functions, such as shutdowns, etc, as these were done by the
full-time staff.

Some background:  when a VM system crashes or is shut down and is brought
back up, you have three options for the type of restart you can do:  Warm
start (look in a save area for volatile data structures from the last
shutdown), Checkpoint (look in your save area for these volatiles, but don't
trust them), or Cold Start (come up as a virgin system, no volatile data
structures such as spool files (mail) or temp disk space).  Well, over the
weekend, the remote campus had suffered a power failure and was without
juice for some period of time.  This was during the ~20 hours a week when
there weren't operators at the home site.  Well, this student operator,
we'll call him Fred, came in and took it upon himself to bring up the
system, quite against rules.  When the prompt for choosing the type of
restart, Fred thought, put his hand on the machine, and said to himself
"Hmm.  It's cold.  Better do a cold start," and restarted the machine wiping
out all the temporary disk space and mail files for the entire system.

>Darin McGrew			mcg...@Eng.Sun.COM
>Affiliation stated for identification purposes only.

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads