|
|
| Author |
Message |
| 25 new of 175 responses total. |
mta
|
|
response 25 of 175:
| 
Nov 3 22:06 UTC 1998 |
Thanks, Mark.
|
shivi
|
|
response 26 of 175:
|
Nov 4 17:57 UTC 1998 |
Dammit how does this work ? I'm too stupid to figure this out.
|
rcurl
|
|
response 27 of 175:
|
Nov 4 19:46 UTC 1998 |
Doesn't look that way....
|
mta
|
|
response 28 of 175:
|
Nov 4 20:58 UTC 1998 |
Welcome to Grex, Shivi! It really does get easier... ;)
|
valerie
|
|
response 29 of 175:
|
Nov 4 23:07 UTC 1998 |
This response has been erased.
|
prometeo
|
|
response 30 of 175:
|
Nov 5 19:08 UTC 1998 |
what's new i'm experimenting with this thing
|
mta
|
|
response 31 of 175:
|
Nov 5 21:51 UTC 1998 |
Welcome, guillermo!
|
tpryan
|
|
response 32 of 175:
|
Nov 6 00:30 UTC 1998 |
Is the auction.cf dead or what?
|
valerie
|
|
response 33 of 175:
|
Nov 7 13:23 UTC 1998 |
This response has been erased.
|
rcurl
|
|
response 34 of 175:
|
Nov 16 17:06 UTC 1998 |
I joined the queue this morning as #95.
|
keesan
|
|
response 35 of 175:
|
Nov 16 17:11 UTC 1998 |
I got busy signals from 8;20 to 8:30 last night, Sunday eve must be one of
those times when the two extra lines actually got some use.
|
remmers
|
|
response 36 of 175:
|
Nov 16 18:55 UTC 1998 |
Nominations for the upcoming Grex Board of Directors election are now
closed. Nominations took place in the Coop conference, item 38
(item:coop,38) The nominees -- for three open slots -- are, in order of
nomination:
Eric Bassey (other)
Jan Wolter (janc)
STeve Andre (steve)
John Remmers (remmers)
Colleen McGee (cmcgee)
Jennifer Kriegel (jiffer)
David Cahill (dpc)
Mark Ziemba (mziemba)
The election will be held online from December 1 through December 15.
|
steve
|
|
response 37 of 175:
|
Nov 19 04:14 UTC 1998 |
Grex had a problem in the passwd database weds night; it was
fixed but Grex didn't allow logins from about 22:50 to 23:08
because of this. Apologies to all who waded through the telnet
queue only to find that you couldn't get on.
|
dpc
|
|
response 38 of 175:
|
Nov 19 14:50 UTC 1998 |
That was an awfully fast fix! Nice!
|
gregb
|
|
response 39 of 175:
|
Nov 19 15:17 UTC 1998 |
Re. 37: Any idea on what happened Tuesday morning? Everytime I tried to
telnet in I kept getting refused connections.
|
valerie
|
|
response 40 of 175:
|
Nov 22 01:40 UTC 1998 |
This response has been erased.
|
tpryan
|
|
response 41 of 175:
|
Nov 22 15:54 UTC 1998 |
Now that you know the system has stability, consider a more
regular routine for re-boot the system, say weekly.
|
shf
|
|
response 42 of 175:
|
Nov 22 19:44 UTC 1998 |
Grex is a very friendly system. It even remembered my daughter's 16th birthday
is today. Astounding!
|
gregb
|
|
response 43 of 175:
|
Nov 23 14:21 UTC 1998 |
Re. 41: I thought with Linux/Unix, you didn't have to do such things.
|
dpc
|
|
response 44 of 175:
|
Nov 23 14:51 UTC 1998 |
That's what I thought, too.
|
mcnally
|
|
response 45 of 175:
|
Nov 23 20:07 UTC 1998 |
re #43, 44: It's less necessary under Unix than with some other operating
systems but running for weeks with hundreds or thousands of users logging
in and running tens of thousands of programs takes its toll on even the most
robust system. I'm telnetted in from another Unix system which has been
up for forty-five days with about 30-40 users logged in at any given time
and it wouldn't surprise me if it ran for another 45 days without problems
but Grex is basically pushing itself pretty hard to operate under typical
conditions day in and day out..
|
mdw
|
|
response 46 of 175:
|
Nov 24 03:32 UTC 1998 |
Actually, it depends a lot on the system as to whether it's necessary.
There is definitely a religion that says "reboot weekly" no matter what.
There are also plenty of Unix systems that stay up weeks, or even years,
under hard usage. The basic issue is the possibility that, somewhere in
the complex of the system, there is a bug that causes the system to get
sick. This bug could have one of 3 major behaviors: it could be a simple
matter of probability; it will happen sooner or later, how often the system
is rebooted doesn't affect it, and once the bug happens, the system needs
to be rebooted. Or, it could be a resource fragmentation problem; if there
is something that needs physically contiguous memory, and something else
that randomly pins pages in memory, then it's possible that with
continued heavy use, the system will tend to evolve in a direction where
the randomly pinned pages are scattered in memory making it impossible
to allocate a large physically contigous chunk. Or, there could be a
resource "leak" problem - the classical one is a system that forgets to
free memory - as it continues to run and allocate memory, it will grow and
grow and grow...
With grex, we don't *know* that the current systems has any of these problems.
We *do* know that previous versions of the system did have stability problems,
and that it was helpful to have weekly preventative reboots. *That* problem
doesn't seem to exist in the current system.
|
valerie
|
|
response 47 of 175:
|
Nov 25 15:29 UTC 1998 |
This response has been erased.
|
atticus
|
|
response 48 of 175:
|
Nov 25 16:13 UTC 1998 |
re #46: I remember reading somewhere that 5ESS telephone switches from
AT&T have an automatic periodic rebooting feature. A switch is an ideal
candidate for problems of category 3 (resource leak) as listed by
Marcus. The number of specific programs are limited, but they are
executed for every phone call that is made. AT&T's technical name for
the periodic reboot is "stochastic rejuvenation", I believe.
|
danr
|
|
response 49 of 175:
|
Nov 25 19:27 UTC 1998 |
Stochastic rejuvenation....I like it!
|
