| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1800.1 | Keep it simple | RLAV::WEGER | Bruce Weger | Thu Oct 20 1988 21:06 | 20 |
|
DO NOT use ^a or anything for any kind of local/forward/backward
character on the terminal server. You will be doomed!
Simply invoke kermit on the VAX and put it in server mode. Period.
The file transfer "type" is negotiated between the requesting kermit
and the server kermit. The VAX kermit (in server mode) will adjust
to whatever you have your Amiga kermit 'xfer mode' set to.
I have NEVER, EVER had a kermit problem on any VAX, PeeCee, UNIX
box or Ami as long as I kept the above rules in mind.
Keep it simple... Don't "set" nothin'...
-bw
P.S. I know its amazing... but don't even worry about tt: lines bits
per character or parity. Kermit will save you ;-)
|
| 1800.2 | Maybe LAT box overruns | STAR::BANKS | In Search of Mediocrity | Fri Oct 21 1988 11:32 | 54 |
| The switch character advice sounds like a good one. I ran into
one more LAT related KERMIT problem back when the company lent me
a 2400 BAUD modem.
With our (now departed) 1200 baud modem, I could download all day,
and upload all day, and generally did so on Saturdays and Sundays.
But, when I plugged in the new modem at double the speed, downloads
got to be real flakey (they'd stop after 20,000-100,000 bytes) and
uploads were nearly impossible.
The common symptom to all of them would be that it'd KERMIT along
for a while, then do its maximum number of retransmissions, failing
with an error message to the effect that it did its maximum
retransmissions and couldn't get an ACK.
Punting out of file transfer mode and back into terminal emulation
mode (generic terms used as I had this happen with both KERMIT and
WeckerTerm version 2.any-I-tried) would put me at a LAT prompt instead
of the expected set of VMS or VMS Kermit prompts.
Now, I'd go out of my way to clear all the forward and break characters
on the LAT port, but it'd still do this. What I eventually discovered
was that the LAT box wasn't doing flow control on the line, or the
flow control wasn't being responded to fast enough. The LAT box
would get frustrated that characters were still coming in too fast,
and just break me back to the LAT prompt. KERMIT wouldn't accept
the LAT prompt as a reasonable ACK, and retransmit N more times
until it gave up.
Now, I don't know which flavors of LAT box this happens with, or
for that matter, which versions of LAT code. I do know that the
problem isn't universal to all LAT servers, because the one I dial
into now doesn't seem to have this problem anymore (and now, I use
KERMITs with big packet sizes, so you'd expect the problem to be
worse, instead of better).
As to what to do if this is your problem: (Assuming that you took
.1's advice and don't set local or forward characters) Before you
connect to the VAX, make sure the LAT port you're using has some
sort of XON-XOFF flow control enabled. I see from your procedure
that you do the opposite. This is bad! The LAT server must be
able to do flow control. Just because the VAX behind it is ready
to accept the characters doesn't mean that the LAT box in front
will have the buffer space.
If it still breaks under those conditions, it may be that
1) I'm describing a problem you're not having
2) The KERMIT program on the VAX is openning the line in some mode
that causes the LAT box to clear or suspend its flow control (PASSALL,
for instance).
3) I'm full of sh*t anyway.
Whatever. At least this is the one that got me a while back.
|
| 1800.3 | Use 2400 baud and Zmodem... | ROMULN::MYEE | Boycott Styrofoam cups, save the ozone. | Fri Oct 21 1988 15:30 | 13 |
|
Re: .2
That's interesting. I have a 2400 baud Scholar modem at home and it
works GREAT on downloads (approx 210 bytes per second with ZMODEM) !!
BUT, I never have any luck with uploads with at 2400 baud (Strange).
The only successful upload I've had was with the Scholar set at
1200 baud mode using XMODEM. I do agree that Kermit is the most
robust file tranfer program, but ZMODEM is so much faster. DOes
anyone have comparative numbers for the different protocal (Kermit,
Xmodem, Ymodem-batch, Zmodem)?
_Mike_
|
| 1800.4 | Zmodem - The only way to fly... | CGFSV2::CADAMS | Clint Adams - Calgary, Canada | Fri Oct 21 1988 16:54 | 24 |
|
When I first started using Zmodem I did some quick benchmarks on the
1200 Baud line. They are as follows:
Kermit(Smokey) Xmodem(Smokey) Zmodem(ProtoComm)
============= ============= ================
62 CPS 96 CPS 113 CPS
Obviously, Kermit is a dog. If you are on the household phone line and
your wife is breathing down your neck the shorter the better for
download times.
Also, Zmodem is supposed to have some fancy error detection/correction
features, but I haven't had any errors reported during downloads so
can't say if that's the case or not.
Xmodem or Ymodem with 1K packets should approach 100 CPS but I haven't
found an emulator for the Amiga end to try that yet.
BTW. Zmodem (at least SZ.EXE) will send multiple files by specifying
them all on the command line.
Regards.... Clint
|
| 1800.5 | BIG packets helps KERMIT a lot | STAR::BANKS | In Search of Mediocrity | Fri Oct 21 1988 17:27 | 48 |
| After working out my LAT problems (solved in the short term by dialing
into something other than a LAT and in the long term with a new
LAT server), I haven't had any more problems downloading or UPloading
at 2400 baud.
However, my fears that KERMIT at 2400 baud wouldn't be twice as
fast as 1200 baud turned out to be very true. Normally, KERMIT
just sends small (around 96 byte) packets, and waits for the other
end to ACK after each one. Obviously, once you start getting into
the higher line speeds, you'll be spending all your time waiting
for the ACK latency.
Two things to do to help this (given that you still want to use
KERMIT):
1) Find a version of KERMIT that does ACK windowing. I don't think
any exist for either the Amiga or VMS.
2) Find a version of KERMIT that does big packets. I think both
WeckerTerm version 2.latest and C-Kermit for the Amiga do this on
the Amiga side. You'll need C-Kermit on VMS as well (as the normal
non-C VAX KERMIT doesn't seem to do big packets). Set the packet
size to something very large (I like 1024 byte packets), and fire
away.
Using solution (2), I've had a lot better luck. Things definitely
move a whole lot faster with 2400 Baud and 1000 byte packets than
they did at 1200 baud with 96 byte packets. Now, most of the latency
I have to suffer is the floppy drive trying to saw the disk in half
on the writes.
While uploading, getting big packets to work is going to be a little
more difficult, though. First off, you'll have to set
TERMINAL/ALTYPAHD, and of course, that has to be done BEFORE you
dial into it (good trick, eh?). Secondly, the ALTYPAHD buffer probably
isn't big enough to handle the 1024 byte packets, so you'll have
to change a SYSGEN parmeter or two and reboot.
The solution I ultimately came to was to put a modem on the serial
port of my workstation. Before I leave for the evening, I shut
down the network on the workstation and enable the modem. From
home, I dial into the workstation (technically not connected to
the network), and do my up/downloading. Since it's my workstation,
I can set the terminal and SYSGEN parameters to whatever I want.
Certainly, this technique requires working for a department which
allows for easy hardware scamming. I realize everyone doesn't have
a spare workstation and scholar sitting on their desks.
|
| 1800.6 | Where? | CIMNET::KYZIVAT | Paul Kyzivat | Fri Oct 21 1988 18:37 | 6 |
| > You'll need C-Kermit on VMS as well (as the normal
> non-C VAX KERMIT doesn't seem to do big packets).
Where can this version of Kermit be found?
Paul
|
| 1800.7 | I can finally upload !!!!!!!! | CAM::ARENDT | DDT - The first and still the best | Sat Oct 22 1988 00:53 | 17 |
|
Ah Gentlemen,
It seems that this discussion has drifted off the main track. However
tonight I don't care. For you see I have (after months of off and
on frustration) finally achieve my goal of uploading files. The
culprit turned out to be the speed of the transfer, I was using
2400, when I slowed down to 1200 'viola' everything worked fine
!! I wish to thank all of you for your help and hints, they really
did help. And I would encourage anyone having trouble with upload
or download to try simply slowing down.
Thank you for your support
Harry who can go to bed now ( it's almost midnight )
|
| 1800.8 | C-KERMIT location (so says HUMAN::KERMIT.NOTES | STAR::BANKS | In Search of Mediocrity | Sun Oct 23 1988 01:15 | 12 |
| Where to get a VMS C-KERMIT?
The latest location seems to be CSCMAS::SYS$KERMIT:CK*.*.
Actually, this will get you sources only (don't know if there's
an .EXE laying around anywhere for this one). It'll also get you
sources for the Unix, Mac and Amiga versions, as well as the VMS
version. You might be able to subset the sources with CKC*.*, CKU*.*,
and CKV*.*.
Still, you'll have to compile it yourself, although I think the
kit contains an .MMS file to aid in building.
|