This item text has been erased.

147 responses total.

This response has been erased.

Not only have we changed over to a new computer, we have upgraded a lot of
software.  Here's a partial list of the changes:

        SunOS 4.1.3_U1  => SunOS 4.1.4
        apache 1.2.1    => apache 1.2.4
        backtalk 0.9.2  => backtalk 0.9.4
        bash 1.14.5     => bash 2.01
        bison 1.24      => bison 1.25
        fileutils 3.12  => fileutils 3.16
        gawk 2.15.6     => gawk 3.0.3
        gcc 2.6.3       => gcc 2.7.2.3
        gdb 4.14        => gdb 4.16
        grep 2.0        => grep 2.1
        less 290        => less 332
        libg++ 2.6.2    => libg++ 2.7.2
        make 3.74       => make 3.76.1
        patch 2.1       => patch 2.5
        perl            => perl 5.004_004
        pine 3.94       => pine 3.96
        screen 3.6.2    => screen 3.7.4
        sh-utils 1.12   => sh-utils 1.16
        sharutils 4.1   => sharutils 4.2
        tar 1.11.8      => tar 1.12
        texinfo 3.6     => texinfo 3.9
        textutils 1.12  => textutils 1.22
        zsh 2.6-beta10  => zsh 3.0.5

If any of these programs don't seem to be working completely right, do let
us know.

I'd like to point out one interesting new command.  The "mps" command is just
like the "ps" command in that it lists running proccesses, but i has one extra
column.  For example, here are some of the proceses listed when I do
"mps -aux"

USER       PID %CPU %MEM   SZ  RSS TT  STAT CPU START  TIME COMMAND
root         0  0.0  0.0    0    0 ?     D    0 02:45   0:31 swapper
popcorn   8274  0.0  0.0   52   60 re    S    0 09:44   0:02 newmail
srw       8752  0.0  0.0  216    0 ?     IW   1 11:24   0:01 -tcsh (tcsh)
srw       8790  0.0  0.0 2088    0 ?     TW   1 11:34   0:11 top
cfadm     8823  0.0  0.3  112  380 rd    S    0 11:42   0:00 bbs
popcorn   8269  0.0  0.0  132    0 re    IW   0 09:44   0:02 -csh (csh)
popcorn   8745  0.0  0.0   68    0 re    TW   0 11:22   0:00 mailx
root      8784  0.0  0.0  308    0 ?     IW   1 11:31   0:00 sendmail: DAA005
popcorn   8842  0.0  0.1   40  128 re    S    1 11:48   0:00 less -dE -q -n
janc      8849  0.0  0.4  224  480 rd    R    1 11:51   0:00 mps -aux
cfadm     8763  0.0  0.0  132    0 re    IW   1 11:26   0:03 bbs

The new column is the "CPU" column which says either 0 or 1.  The 4/670 is
a dual-processor system (with each processor being individually much faster
than the old system's CPU was) and this tells you which CPU that particular
process is running on.  So Valeries's login shell and mail program are
running on processor zero, but her pager is running on processor one.
Cute, huh?

IWLTA that this is posted from BackTalk on the 670.  If you can read 
this, it worked. ;)

Thanks from me to all of you.  I know how much time and effort you put into
making this work, and although I don't understand all of what you did, the
system is up and running and is quicker then a bunny.  I send you all virtual
hugs.

Thanks also to Jared Mauch and Mike McNally, both of whom leant invaluable
assistance in getting the 4/670 up and running.

Good job and hats off to all . . .

Ditto!

Good work....the system is noticeably faster. It seems a lot of accumulated
good ideas have been implemented all at once. Congratulations!

Excellant job and thank you! :)

We think that each processor should run at several times the speed of 
the previous computer. Because of limitations in the operating system, 
the second processor may not be able to be used at 100%, but it will 
still help to speed things up. In addition, the new architecture should 
be much faster at switching "contexts", which is something done a lot 
under Grex's load. So we don't really know how much faster the new 
machine will turn out to be.

The mail spool hierarchy will also generate a component to 
the performance boost.

We were very seriously limited by our processor speed before this 
change. For all we know, we may still be limited by it, but only time, 
and the heavy load generated during the week, will allow us to know for 
certain.

Right now there are 42 users, and the load average is 1.73.  Looks great.
Thanks everyone for all the work!

Thanks, all!  

Many thanks to staff, whose hard work seems to finally be paying off.  As I've
already mentioned, performance is already noticeably improved.

  First, let me say that I'm happy that my modest contributions have helped
  bring about this successful upgrade.  Second, allow me to congratulate
  those who are chiefly responsible for what seems to have been a very well
  planned and smoothly executed switch-over.

  For those who don't have a great deal of experience judging such events,
  the accomplishment of successfully switching over so many things at once
  and having everything work smoothly afterwards should not be underestimated.
  Take it from me, who has done this for a living for a number of years, 
  making major changes to an established service, particularly making a number
  of simultaneous changes, takes careful planning and a good deal of meticulous
  coordination.  Once again, Grex's staff and core group of volunteers have
  accomplished wonders..  Good job!!

(minor nit, but Grex's staff *is* entirely a dedicated core of volunteers)

For some reason, the biggest speed improvement I've seen thus far is on the
 !man command.  Wonder why that is....

I can confirm that the checklist of things to move and do was long, and 
some staffers spent a lot of time going over everything before it came 
up for real earlier today. Still the glitches can be counted on only one 
hand. I'm a little startled at that. 

 re #16:  by "Grex's staff and core group of volunteers" I wasn't trying
          to suggest that Grex's staff are not volunteers but that there
          are volunteers who were much involved in this and other projects
          who are not staff.

Indeed, I've been quite surprised at how smoothly everything is going (Pine
took an unprecedentedely short one second to load).  I haven't noticed any
problems yet.

        Have the manual pages for all the updated software been likewise
        updated?

        Thanks and great job!  I can't remember the last time I saw Grex
        move this fast, especially with 64 people on! 

  Given the success of the upgrade, would it be pushing things if I were
  to lobby for the addition of the nvi editor?  It's basically an updating
  of vi for modern times by some of the Berkeley CSRG people that keeps
  vi's chief benefit of being lean and mean while still adding useful
  features like 'unlimited' line lengths, editing of binary files, multiple
  undo levels, multiple 'windows', etc..  It shares a lot with vim,
  which is already compiled here on Grex, but I prefer nvi because it
  sticks a little closer to the original vi design with the features it
  has added. 

  I'd be glad to compile it if none of the staff have the time, though I
  can understand why that's not a practice to be generally encouraged,
  anyway it's an easy build..  

Backtalk is smoking with the new upgrade. It's very fast now.

   Mike, bring it over and compile it.  As a user of Grex you certainly
have that right automatically.  I've heard of nvi but haven't played
with it.  If we can reasonably run it, I think we should.  It's one of
those things, which while probably being a resource hog, won't be used
by that many at once.  We can at least test it out here.

Backtalk is running very quickly indeed.  For us WWW users, this was a 
very significant upgrade.

Thanks!

This response has been erased.

It's no more of a resource hog than vi is, I don't think.  I have no problem
with it.

Oh, I had entered special thanks for Mike and Jared, but I forgot Rob Argy.
Rob did most of the shopping for parts for the new machine and helped put it
all together.

Let me add my "THANKS!" to those responsible for making an apparently seamless
upgrade to grex.  Right now it is "screaming" right along, and the browse
command is even speedy!  :-)

Thanx to all!  Excellent work--Grex is usable again!

Wow--ttyuse even works!  I'm dialing in, and wound up on t8.
Are the t* ttys the dialins, or is it random?

The  radio  conference has been restarted. Join radio to find a whole
new space in which to discuss any aspects of radio, from amateur radio
to z... What begins with z? (The form radio cf. is archived as oldradio.)

pty assignments by telnetd, rlogind, & sshd are random.

Mike, we don't discourage people from bringng over packages and 
compiling them. We merely ask them to ask staff first. There are two 
reasons we ask that:
(1) In many cases the application has no chance of working on grex,
    or no chance of working except for members.
(2) In many cases the application is already installed here.

Reasonable requests for staff to build the package and install it are 
usually honored, but this varies as it depends on the nature of the 
program and the availablility of staff time and CPU time. nvi looks like 
something we would do upon request anyway.

(Well, CPU time is more plentiful than staff time at the moment. )

Wow!  Thanks for all of the hard work!  I can actually see what I'm typing
AS I type it...  ;-)

Thanks to all staff members and everyone who contributed (especially Grex
current and past members) for your contribution to the current system!  It's
_much_ faster!  50 users and a load average of < 1....mightily impressive!:)
No major lag bouts to me here in Aussie yet either:)  Thanks again to you all!

        What is load average and how does it affect the speed?

The load average is related to the number of items waiting to be
processed by the System.  The lower, the better.

        I'll add my thank you here.  Good work, with sucess at the end.
You all musta done a lot of testing in the past weeks & months.

- Backtalk version 1.3.30 - Copyright 1996-2006, Jan Wolter and Steve Weiss