Subject: Re: sup-server brokenness...
To: Mike Long <>
From: John F. Woods <>
List: current-users
Date: 10/10/1995 16:00:51
> 1) (aka is always busy.  Has
> *anyone* been able to sup successfully from this machine?

Yes.  It took a long time, but sup is good at waiting, so I let it do it.

I assume that is being hammered by several dozen people
trying to sup, and since its file dates are all different from sun-lamp's
(presumably sun-lamp's crash made it impossible to restore the original
file times), those lucky few who succeed get to watch sup take several hours
to work.  ( also appeared to crash once during a sup, which
didn't help.)  At roughly three successes per day per allowed simultaneous
connection, it'll take a long time to empty a queue of several dozen...

> Problems like this one probably do more to drive users away than
> install tools, politics, or anything else.

It's real hard to recover from a disk crash on an active system like a sup
server.  It's even harder when it doesn't have a paid attendant to do daily

Perhaps individuals should refrain from running sup until some of the mirrors
have caught up?  (And those mirrors that are using ""
instead of "" should be updated pronto.)

Once the mirrors catch up, what are the issues involved in having users
alternate between different sup servers?  I.e. would it be practical to
spread the load by having "" resolve to several shuffled
addresses?  (Of course, then the mirrors have to use ""
or something like that.)