8/4 With both kais and mehlhaff's archive out of action (why?), can someone
please point to another public access archive? Sometimes motd even
contain useful information, like the script posted early here that I
did not save. (I also cannot run it for the lack of disk space.)
\_ my motd archiver is back online. IT requires manual restarting when
soda reboots. Not a problem when soda is stable... -ERic (mehlhaff)
\_ kchang seems like a smart guy who does stupid and annoying
things, why didn't he run a script on soda that diff'd the motd
then transferered the diff bits back to him at ucla?
fingering 60 times a minute seems like a very inelegant
non kchang perl hack solution.
\_ How the hell could finger be considered a DoS attack?! Was this
just done for personal vendetta reasons against him?
\_ back when the net was young and the hills were green,
system administrators logged tcp connections like rlogin
and finger and who and stat and ftp. now no one cares anymore,
all the traffic is http, but the EECS admins probably
have the same logging scripts in place from the early 90s.
i remember some soda dude who went off to stanford would
finger the account of his ex-gf on soda like once every
5 minutes, for hours, at random times, all day every day.
it stuck out in the logs because it was always the same
domain name, to the same account, all the time. i emailed
the victim and she probably emailed the dude and he stopped.
i can see how the EECS admins would notice kchang fingering
the motd every 10 seconds.
\_ The above seems to refer to kais' statement. So is kchang
squished again because he runs a world accessible motd archive?
\_ no
\_ Considering the amount of censoring he does on the entries
before they are categorized and posted, I find it less than
useful as an "archive."
\_ He was fingering Soda through EECS more than once a second.
It filled up the EECS network logs and made them unhappy.
Read the last meeting minues and following discussing on the
newsgroup.
\_ can't they just ask him to finger less?
\_ they could but kchang is unpopular enough that no one
at the csua would suggest the obvious common sense
solution. if everyone lost an account on the first
strike, then we all would've been squished by now.
kchang is special. it was an unpopularity squishing
like all of them.
\_ "all of them"... I think you know naught of what
you speak. Longstanding CSUA (perhaps unwritten)
policy is "if you get your account back, don't
fuck it up." That said, I can't speak to why they
actually squished kchang, but comparing a second
squish to a first is apples to oranges. --scotsman
\_ if it isn't written, it isn't policy. if you want
a policy that says a second squishing has a lower
standard than the first or that post-restoration
from a first squishing a person is on probation
of some sort, then write the policy and publish
it. otherwise, yes, it is apples to apples. orgs
are properly run by written policy, not whim.
\_ ok, I checked the newsgroup. from the discussion,
it seems that they were not about to squish him,
and even made fun of tom when he tried to chime
in and get kchang squished. so what changed their
minds in the end?
\_ "they" != politburo
\_ warned, deactivated in 98. reactivated and warned
to behave. deactivated again.
\_ it also caused problems for EECS because soda makes ident
requests to remote systems that access soda's finger server.
imagine the EECS webserver getting pummeled by multiple
identd requests every second.
\_ Here, now you can run your own private script. Don't forget to
mkdir /csua/tmp/username. The next thing that can be done to this
is to prevent check-ins if the motd is about like 100 KB (someone
puts binary junk in it):
#!/bin/sh
cd /csua/tmp/$USER
while [ 1 ]
do
co -q -f -l motd
cp /etc/motd.public motd
ci -q -m_ motd
sleep 120
done
\_ Or make it check the output of `file /etc/motd.public` ro make
sure it says ASCII.
\_ BAD IDEA. If everyone on Soda did this, the space would fill
up in no time.
\_ I run my own motd archive. I once posted the path to it. Check
the archives.
\_ Thanks for recursing. |