|
|
| Author |
Message |
| 25 new of 170 responses total. |
mdw
|
|
response 50 of 170:
|
Oct 29 21:27 UTC 2001 |
"Would have"? Network problems can happen anywhere along your path, not
just at grex's end.
|
carson
|
|
response 51 of 170:
|
Oct 30 00:24 UTC 2001 |
Allow me to clarify: connections from Chinet to Grex were refused.
Connections from Longyear to Grex were either refused or abruptly
closed. At least one of the closures cited Grex's IP address as the
source, while others simply indicated an ambiguous "remote host."
Addendum: all seems well now.
|
mdw
|
|
response 52 of 170:
|
Oct 30 04:20 UTC 2001 |
Sounds like it could be a firewall issue, but it's hard to tell. A
refused connection is most likely either a firewall or inetd not running
on grex, perhaps because grex isn't in multi-user mode (ie, backups). A
traceroute and captured packets might help to distinguish the two cases.
A connection that abruptly fails sounds like something died at the
remote end, like telnetd. A connection that hangs for a while, then
disconnections, could be due either to a network routing glitch of some
sort (including the DSL link at grex's end going down), or grex itself
crashing. A connection that is made to grex, then rejected before any
mesage appears, may be due to the tcp wrappers. A connection that is
made, a login prompt appears, then the connection is broken before the
loginid can be entered, is probably due to a conflict between your
telnet client, and grex's telnetd: grex will only accept "valid" baud
rates for the connection, in which case, "stty 9600" before the telnet
connection to grex may fix things.
In most of these cases, the error message you'd see has little to do
with the cause. A timeout due to a network problem may be reported as
"the remote server disconnected you" when there's in fact no way for
your local end to know what the remote server was thinking.
|
gull
|
|
response 53 of 170:
|
Oct 30 15:51 UTC 2001 |
ssh also times out much more quickly when there are network problems
than telnet does. This sometimes makes ssh unusable when the network
is laggy.
|
tsty
|
|
response 54 of 170:
|
Nov 2 12:27 UTC 2001 |
this was pretty weird about 2 minutes ago ... :
mkids too small
#297 of 302: by JP2 * GREX * AND YOU! (jp2) on Wed, Oct 31, 2001 (10:48):
#298 of 302: by Sindi Keesan (keesan) on Wed, Oct 31, 2001 (15:40):
(Fixed item 92 flags 5c->9e mtime 0)
mkids too small
(Fixed item 92 rcnt 303 mtime 0)
mkids too small
(Fixed item 92 flags be->9e mtime 0)
#299 of 302: by Megan Heberlein (eeyore) on Thu, Nov 1, 2001 (00:26):
mkids too small
mkids too small
(Fixed item 92 flags 15e->5c mtime 5c)
mkids too small
(Fixed item 92 flags 5c->9e mtime 0)
mkids too small
(Fixed item 92 rcnt 303 mtime 0)
mkids too small
(Fixed item 92 flags be->9e mtime 0)
#300 of 302: by Sindi Keesan (keesan) on Thu, Nov 1, 2001 (09:13):
mkids too small
mkids too small
(Fixed item 92 flags 15e->5c mtime 5c)
mkids too small
(Fixed item 92 flags 5c->9e mtime 0)
mkids too small
(Fixed item 92 rcnt 303 mtime 0)
mkids too small
(Fixed item 92 flags be->9e mtime 0)
#301 of 302: by That Anne Person (mooncat) on Thu, Nov 1, 2001 (10:49):
mkids too small
mkids too small
(Fixed item 92 flags 15e->5c mtime 5c)
mkids too small
(Fixed item 92 flags 5c->9e mtime 0)
mkids too small
(Fixed item 92 rcnt 303 mtime 0)
mkids too small
(Fixed item 92 flags be->9e mtime 0)
#302 of 302: by Megan Heberlein (eeyore) on Thu, Nov 1, 2001 (23:19):
mkids too small
mkids too small
(Fixed item 92 flags 15e->5c mtime 5c)
Item 92 was killed!
Item 92 was killed!
Ok: q
You are now leaving Fall agora.
haven't checked yet to see if #92 was, in fact, killed.
|
janc
|
|
response 55 of 170:
|
Nov 2 15:15 UTC 2001 |
I tried to figure out what all that might mean, and failed. "mkids too small"
has to do with running out of file descriptors. Can't imagine why that would
be happening for a whole bunch of people. I think the "Fixed item" stuff
refers to fixing the flags in the conference sum file. These tell if the
item is linked, frozen, etc, but normally there is a more reliable information
about this attached to the item. Why it would make the same series of fixes
for multiple users is beyond me. Maybe it really has to do with fixing
participation files, not sum files (doubt it). Also, I've never entirely
figured out what all these flags mean, and the ones in the example above are
among the weirdest I've seen.
However, one case where you often get this kind of stuff is when someone with
root or cfadm powers has been editing the raw item.
|
aruba
|
|
response 56 of 170:
|
Nov 2 16:26 UTC 2001 |
I always get the "mkids too small" message after I've been entering a lot of
items or responses without exiting Picospan.
|
eeyore
|
|
response 57 of 170:
|
Nov 2 16:30 UTC 2001 |
Well, for what it's worth, I never saw any of those messages, and I'm sure
that I posted something else....damned if I know what, though..,..
|
mdw
|
|
response 58 of 170:
|
Nov 2 18:14 UTC 2001 |
I think I have fixes for the mkids thing, I'll try to get around to
installing it on grex someday soonish...
|
tsty
|
|
response 59 of 170:
|
Nov 3 06:42 UTC 2001 |
kewl, whatever ... just teh messenger ...
|
davel
|
|
response 60 of 170:
|
Nov 3 15:02 UTC 2001 |
Grex just unceremoniously dumped my dialup connection again. <sigh>
|
jp2
|
|
response 61 of 170:
|
Nov 7 20:22 UTC 2001 |
This response has been erased.
|
jhudson
|
|
response 62 of 170:
|
Nov 8 01:43 UTC 2001 |
It is in there now.
|
keesan
|
|
response 63 of 170:
|
Nov 10 18:16 UTC 2001 |
I would like to report that for the first time ever I was able to download
a file of more than 12 packets using Kermit - are the new modems installed
or is it something to do with low usage on this particular Saturday?
Usually it crashes at 12 packets and I got to 18.
|
keesan
|
|
response 64 of 170:
|
Nov 11 22:33 UTC 2001 |
The new lynx 'can't access startfile' when I typed lynx plus URL.
It looks for a local file. The second to newest lynx (newlynx)
works fine. The newest one is looking in local file not the web.
Perhaps the lynx.ini file needs adjusting?
|
blaise
|
|
response 65 of 170:
|
Nov 12 03:05 UTC 2001 |
Sindi, does it make a difference if you start the URL with http://?
|
keesan
|
|
response 66 of 170:
|
Nov 12 14:43 UTC 2001 |
Yes, if I type in the http:// first the error message is not 'Can't access
startfile' but 'Unable to connect to remote host' and then a bit later
lynx. Can't access startfile http://..... (instead of the bit about local
file). Try it yourself if you can telnet or dial to grex. I do the same
thing with newlynx and newlynx works as expected (2.8.2). Proxy problems?
I am dialed in but telnetting produced the same problems yesterday.
|
blaise
|
|
response 67 of 170:
|
Nov 12 15:01 UTC 2001 |
Yes, I see what you mean. Using http:// works, but using a bare hostname
doesn't.
|
keesan
|
|
response 68 of 170:
|
Nov 12 17:00 UTC 2001 |
Neither worked for me, dialed in. At 9:42.
|
keesan
|
|
response 69 of 170:
|
Nov 12 17:28 UTC 2001 |
I just tried again and lynx http://www.paypal.com does work. Jan says it
will take some clever work to get lynx www.paypal.com to behave as expected.
|
janc
|
|
response 70 of 170:
|
Nov 12 17:31 UTC 2001 |
I haven't had any problems when I use the http: prefix. What site are you
trying to access Sindi?
The bug with the bare names is because lynx wants to use the system five
version of curses, and so accidentally also links in the system five version
of the DNS host lookup function, which doesn't work on Grex. I haven't
figured out how to resolve this one yet.
|
janc
|
|
response 71 of 170:
|
Nov 12 17:32 UTC 2001 |
Sindi slipped in.
|
tpryan
|
|
response 72 of 170:
|
Nov 12 20:38 UTC 2001 |
Oops, I put my observations in the last item. I'll expect
responses to my lynx observations here.
|
keesan
|
|
response 73 of 170:
|
Nov 13 00:33 UTC 2001 |
The following work: lynx http://www.cyberspace.org
newlynx http://www.cyberspace.org newlynx www.grex.org newlynx
http://www.grex.org
But lynx http://www.grex.org gives me an error message:
Alert! Unable to connect to remote host....
Can't access startfile http://www.grex.org
Why is this?
|
janc
|
|
response 74 of 170:
|
Nov 13 03:34 UTC 2001 |
I've installed a new version of lynx that seems to do DNS right. It also uses
ncurses instead of curses to do screen management (believe it or not, changing
the screen management library was the easiest way to make it do DNS lookups
correctly). Ncurses is supposed to be able to do color, if the termcap
supports it. I'm not sure if they do. Note that ncurses and thus lynx uses
different termcap definitions than every other program on Grex. I'm hoping
this won't matter to most people. I'm hoping we can move to OpenBSD soon.
|