response 75 of 196:

Jul 16 21:47 UTC 1996

Yeah, maybe 25 or 30 slots in the que is reasonable...the percentage of
people who are going to stay connected if they are #785 or something is
probably very small.

Jared had a similar setup on nether.net for a while...but it seems like it
must have caused more problems than it solved, because he took it off months
ago and returned to "all ports are busy"

response 77 of 196:

Jul 17 09:07 UTC 1996

First off, I'd have to say that I am *so* happy that we have the queue in
place now.  I think that I'll actually be able to get online sometimes now!

Second, *please* don't make the timeout on the login: prompt any shorter --
especially 10 seconds!!!  Last night, there were periods where netlag from
Grex to ICnet was running up to 20 seconds.  *No one* would be able to log
in if the thing were set to 10 seconds, and they probably wouldn't be able
to log on if it were set to 30 seconds, although that would be a unique way
of eliminating netlag above a certain threshold.  8^)  By the way, I
redirected the input from that ping that I started at 12:45am EDT to ~2:25am
EDT to a file called /home/nephi/ping if anyone wants to take a look at it
and verify what I say -- or you can just tail it for the summary.  (And yes,
we *do* have packet loss between us and ICnet on a regular basis.)

Third, please *don't* limit the size of the queue (beyond some absurd limit,
where Grex performance will start to die)!  If someone connects, sees that
he has 150 people in front of him and wants to wait 45 minutes to connect,
what's wrong with him sitting in line?  Isn't the whole *point* to get people
to stop attack telneting?  

Fourth, I personally think that it would be really neat to display a copy of
the MOTD in the queue.  I can see some possible problems/objections with this,
though, so I'm just wondering how other people feel about it.  

        Waiting for a free port (? for help)
        ...66

does seem a *bit* austere.  

Fifth, having a wait queue will most likely change our demographics in favor
of people who can plan their lives out 45 minutes in advance.  Party will
probably end up getting used less, as that is a more "spur of the moment" type
of program . . . and things like BBS, email, and lynx will probably end up
getting used more.  We will probably start favoring older, more patient, users
over younger, more impulsive, users.  The newuser program seems like it will
get much less use under this system.  I don't know very many people at all
who would wait between 15 and 45 minutes to log in to some mystery system that
they know nothing (or very little) about.  A "natural" way for us to curb our
future growth?  

Sixth, I *love* Marcus to death for writing this!  If he were here he'd be
getting major hugs right now . . . 

response 79 of 196:

Jul 17 16:41 UTC 1996

     This morning I was #36 in line.  It took a half hour to get a login 
prompt and the minute I did, the connection closed.  Tried again.  This 
time was #31, took twenty minutes to get a login screen and the minute I 
did, "connection closed"  So it took more than an hour to get on grex 
this morning, which since I was multi-tasking wasnt any great 
inconveninece, but is still  ridiculous.

The countdown que is patently unfair because it gives the advantage to 
the dial-ins.  When you dial-in you can just put on automatic re-dial and 
take your chances until you get an answer.  Almost exactly like the old 
"all ports are busy" telnet setup.  Dial-in and you do not have ot wait 
in line.  Telnet and you do.  It occurs to me that this may be why much 
of staff and the board want this setup.  They more often than not 
dial-in.  There are plenty of members out there who dont like the telnet 
ports because they overload grex and involved mostly non-local people.  
So this deal is like an equalizer!

It seems to me that this is going to strongly discourage folks from 
telnetting and will drive people away from here.  As was pointed out, it 
was incorrect of staff to assume that telnetting patterns and practices 
follow praticularly consistent patterns.  In fact most didnt telnet 
bomb.  Telnetting and getting "all ports are busy" and then doing so 
again thirty seconds later and at thirty second intervals untilyou get a 
login is not a "telnet bomb".  It is automatic redial and there is 
nothing wrong with it!

I dont define a problem as such un;less a lot of people are complaining.  
I never saw that many complaints about the "all ports are busy"  Maybe 
once in a while if the system was particuarly slow or something, but 
generally people werent complaining.  I maintain that it is wrong for 
staff to implenent such changes without a board vote.  Not just 
discussion, but a vote (and not a stupid "not to object" vote)  I think 
this should be left up for a week and then taken down and discussed and 
voted upon based on the observations made during the week it was 
experimented with.  This is fair.  But since staff wants this change, 
perhaps the idea of of fairness isnt the point.  What is the point of 
discussion if the decision has already been made?

Be fair about this. Put it on the agenda at the next board meeting, and 
accept this as a trial run for a few days.  I really dont think the 
ramifications of this change have been fully gone into.

response 81 of 196:

Jul 17 16:58 UTC 1996

Considering many of the telnets are from people local to AA, I would
suggest that the dial-ins DON'T have an advantage.  If they did, 
nobody from AA and environs would be telnetting.
There are some 62 telnet slots.
There ae some 11 dial-in slots.
Give that those 62 have to spread out over the world, and those 11 have to
spread out over a few dozen miles, I should think that the fact that any AA
folk at all are using the telnet indicates that at least those people prefer
standing in a line that might take 20 minutes  to attack dialing.

(Granted, some of the AA folks are telnetting from sites which don't allow
dialins,  but I'd guess taht's the minority, at least during the summer.)

response 83 of 196:

Jul 17 17:12 UTC 1996

To elaborate, reasons why ths countdown que is unreasonable and unfair:

1. It discriminates against people who don ot have the option of 
dialing in.  SCG lives in Ann Arbor.  If he is #37, he can hang up his 
ISP and dial direct, re-dial and get in much ch faster.

2.  many people on grex pay for t heir ISP's by the hour, and cannot 
afford to waste a half hour of metered time waiting on grex's countdown 
que.  This will drive thedm away.

3.  Many get on from public places like the library and have 30 minute or 
hour time limits, so many times a countdown excludes them as well.

4.  many people do not have the ability to multi=-task, and asking them 
to tie up their computers and their phones for a half hour just so grex 
can countdown, is unreasonable. These are people who need to be able to 
re-connect periodically, in order to mak,e it practicaql.

5. If the net lag is bad, coutndowns can be unbearably slow.

6.  The problem of connections closing once the login prompt appears.

7.Simple personal freedom.  the choice of how, by what method and how 
often one telnets should be left up the individual.  Grex staff shuld not 
be dictating telnetting pattersns to its users.  If staff cannot dicatate 
how many times a user can hit automatic re-dial when dialing in, why 
should they exert the authority to tell users how often they should 
telnet.  How would the dial-in folks like it if staff put in a setup 
where instead of a busy signal they got an anser but were kicked into a 
waiting room instead of getting a login prompt.  ? their telephones would 
be tied up and they'd be pissed.  They'd say it is an imposition to force 
me to wait in line.  Let me call back!  

Give the telnetting users the same courtesy.  If everyone was local and 
everyone had unlimited hours ISPs and everyone could multi-task and come 
back to their grex window, maybe this would be a good idea.  But the fact 
is that everyone does NOT have the same capablitiies, and this 
discrinites against many who do not have the best computer setups.

response 85 of 196:

Jul 17 17:35 UTC 1996

(#82 and #83 slipped in.)

response 87 of 196:

Jul 17 21:34 UTC 1996

Richard, you mention personal freedom.
Okay.
With this, you have the freedom to wait in the queue or com back when it's
less busy.  I've logged on twice today.  guess what?  No queue.  Not one
second.  I log on during slower times, or try to... I also log on during
peaks.  When I log on during peaks, I expect a wait.
Now, without this, you don't have an option.  You have to come back later.

This way, choices:  stand in line or don't.
The other way, choices:  don't.
 How is this infringing on personal freedoms, PC boy?

response 89 of 196:

Jul 17 22:37 UTC 1996

(I've *got* to stop Grexxing when I'm crabby . . . 8^)

response 91 of 196:

Jul 17 23:15 UTC 1996

I don't think the Board ever formally voted on this. It may have
discussed it informally as part of a technical committee report,
but I don't know as I've missed a few meetings since the end of my
Board term. In any case, knowing the people on the board, I can't
believe that the Board response would have been anything other
than "Sure, let's give it a try if somebody's willing to put in the
work to implement it."

I'm still plagued by the "immediate disconnect at login prompt"
problem. Tried twice just now -- first time, I had the problem,
second time, I didn't and was able to log in.

response 93 of 196:

Jul 17 23:22 UTC 1996

maybe the idea is to get thebest of both worlds...set the que max at 10
This way more peoplewill get"all ports are busy"but at the sametime 
fewer will be waiting in the que.

also why has this change re-pointed "quit" and "exit" to the login prompt?
I doubt many who type those commands want to log in again.

response 95 of 196:

Jul 17 23:43 UTC 1996

To add to #94: The more specific you can be about what time it happened, the
better.  Also, we're interested in knowing what software you were using to
telnet to Grex, and what kind of machine you're coming from, to see if there's
a pattern.

response 97 of 196:

Jul 18 00:56 UTC 1996

   One of the reasons the queue is more active and the waits are longer 
is because this is being tested while M-Net is down instead of under 
normal circumstances.

   If I were a newuser telnetting in, I would be more likely to wait in 
line than telnet in repeatedly.  After the first dozen or so times I 
received "all ports in use", I would assume the system was screwed up and 
leave.  With a queue, at least I know it's working.
   With places like the library and Cybercafes, the queue is still 
better.  With the old telnetd, you could either crush Grex with telnets 
or leave and try again later.  Now, you can join the queue and instantly 
know if sticking around is worth your time or if you should just leave.

response 99 of 196:

Jul 18 15:36 UTC 1996

One point tht keeps sticking in my head...
Some people get here through Gopher.
People who go to the same board everyday memorize the keystrokes.
Keystrokes can be mistyped, and if done too quickly, menus can be skipped.
In the bad old days when I used Gopher, I mistyped now and then,
and was frustrated to find that my login wasn't working, until I realized I
was trying to log in to a foreign system.
?That said, shouldn't the queue message at *least* identify the system?
I mean, just "Waiting for Grex..." would be fine.