henning@thor.acc.stolaf.edu (Mark D. Henning) (10/09/89)
Hi, first of all, I would like to say how pleased I am with the Dnet-net: program team. I do, however have a problem. I am able to get the net to work properly at speeds of 19200 and lower. I am also able to get the net to work for a limited time at 31250. Dnet connects fterm at 57600, but netmount will not function. The problem occurs when I have mounted a device on the remote machine and then call for a directory. It will sometimes finish, but more often than not, it will lock the network (and cli) up. The network is then dead to the other CLI's as well. From these facts, I conclude that I either have something incorrect in my dnet config files, or there is a bug in either dnet or net: (I would assume that dnet would request mangled blocks over, if the checksums were bad.) I open dnet over a 300 foot+ 3 wire null-modem with the command: run dnet -X -8 -bxxxxx -h0 I see two solutions... one hardware, one software (or configuration). The Hardware solution would be to boost the signal (after all rs232 is rated for 50 feet and my run is at least 6 times that), translate the signal to something rated for longer runs (rs423?!? the mac standard... I don't have a manual nearby, so I'm not sure of the exact rs number), or even better... translate this new mac compatible signal to "PHONENET" (mac's "listen in" network system) which would allow for multiple node networking in the future, as well as pulls of nearly 1000 feet. The software solution involves feedback from those more experienced with dnet. I hope that My hardware musings are wasted time. If anyone has any advice, please email me the answer. Thanks in advance -henning -- ------------------------------------------------------------ All opinions are my own, as are the spellings any deviation from the norm is just an effect of late nights and early mornings, with little sleep between.
aaron@madnix.UUCP (Aaron Avery) (10/11/89)
In article <7222@thor.acc.stolaf.edu> henning@thor.acc.stolaf.edu (Mark D. Henning) writes: >I am able to get the net to work properly at speeds of 19200 and >lower. I am also able to get the net to work for a limited time at >31250. Dnet connects fterm at 57600, but netmount will not function. Are you doing this on the built-in Amiga serial port? That port can't reliably receive at speeds greater than 19,200 bps, when using the serial.device. Sends are ok, but they generally don't go out at full speed. My suggestion for speed is to use Matt's new parallel network stuff, or get an expansion serial port which can reliably send AND receive at speeds of 57,600 bps and greater. We produce such a board. - Aaron -- Aaron Avery, ASDG Inc. "A mime is a terrible thing to waste." -- Robin Williams ARPA: madnix!aaron@cs.wisc.edu {uunet|ncoast}!marque! UUCP: {harvard|rutgers|ucbvax}!uwvax!astroatc!nicmad!madnix!aaron
walker@sas.UUCP (Doug Walker) (10/19/89)
In article <7222@thor.acc.stolaf.edu> henning@thor.acc.stolaf.edu (Mark D. Henning) writes: >I am able to get the net to work properly at speeds of 19200 and >lower. I am also able to get the net to work for a limited time at >31250. Dnet connects fterm at 57600, but netmount will not function. This is exactly the behavior I have seen, and I mentioned it as such in my documentation for NET:. >From these facts, I conclude that I either have something incorrect in >my dnet config files, or there is a bug in either dnet or net: (I >would assume that dnet would request mangled blocks over, if the >checksums were bad.) I don't know what DNET is seeing, but from the network's point of view I see: 1) The handler submits a packet to DNET for writing; it gets a successful return code. 2) The server receives and processes the packet, then writes the result back via DNET again. Successful return code. 3) The handler never receives the packet. It's dead in the water. Maybe I'm misusing DNET, I don't know. The source code is posted, so y'all can second-guess me. --Doug
dillon@POSTGRES.BERKELEY.EDU (Matt Dillon) (10/19/89)
:>I am able to get the net to work properly at speeds of 19200 and
:>lower. I am also able to get the net to work for a limited time at
:>31250. Dnet connects fterm at 57600, but netmount will not function.
:This is exactly the behavior I have seen, and I mentioned it as such
:in my documentation for NET:.
The Amiga gets overrun errors on anything above 19.2KBaud
when you use serial.device ... custom midi drivers can go faster,
and third-party add on serial boards which have on-board FIFOs
can usually handle any rate.
But with the internal serial port, anything above 19.2KBaud
means the Amiga's rx interrupt can't handle the rate, gets
overruns (a MASSIVE number of overruns), the serial.device goes
nuts and sometimes repeats parts of older buffers, and THIS
causes DNet to go kapoof.
Realistically, 9600 or 19.2KBaud are as fast as you should run
it. Under 1.4 you ought to be able to run faster as many OS bugs
(Disabling for too long a time) will be fixed.
-Matt