plip tx/rx timeouts during file transfer

plip tx/rx timeouts during file transfer

Post by David Efflan » Thu, 18 Jun 1998 04:00:00





double with 1K chip cache enabled).

With a 38.4K serial ppp I was getting 70 ping and 1.7K/sec transfer
rate due to many errors with the old UART.  With plip I get a 5 ping
and 40K/sec transfer rate.  Plip module set for a specific port even
works alongside auto lp module on another port.

But during file transfer, I get frequent messages flashing to the
screen like 'receive timeout (4,c7)' and 'transmit timeout (4,87)'
while saving files on the slow machine.  The files seem to transfer
intact, but the console messages mess up the screen display of ncftp.
Does plip have any flow control?  Can these messages be selectively
suppressed without suppressing other errors?

I don't seem to have the 'plipconfig' program mentioned in the PLIP
mini HOWTO, but I did notice some variables in the module I could try
experimenting with:

PLIP_DELAY_UNIT
PLIP_TRIGGER_WAIT
PLIP_NIBBLE_WAIT

Possibly increasing the delay unit on the fast machine would make all
its plip timers run longer.

David Efflandt/Elgin, IL USA

 
 
 

plip tx/rx timeouts during file transfer

Post by David Efflan » Fri, 19 Jun 1998 04:00:00


Never mind.  I tried setting PLIP_DELAY_UNIT to 5 (instead of 1) in
plip.c of the fast machine and remade modules.  This eliminated the
timeout errors while transferring files to the slow machine.  No

David Efflandt/Elgin, IL USA


 
 
 

1. Number of tx vs. rx packets for ftp transfer

I have a 2.2.6 box that I used to ftp navigator 4.6 from
ftp.netscape.com via cable modem last night.  After the
transfer was done (~9.5 Meg), I noticed that the eth0
interface had about 10000 Tx packets, and 6000+ Rx packets.
The interface was just brought up prior to the transfer
so all of these packets were due to the ftp transfer.

My question is, doesn't that seem a bit high for Rx packets?
I actually have no idea what it should be, but I was sort
of taken aback.  It seems strange that I need to send one
packet for every 2 that I get, especially for a simple
download.

The reason I'm concerned is cable upstream is potentially
bandwidth capped, and apparently using upstream bandwidth
can degrade downstream bandwidth, and I've got absolutely
*abysmal* performance with my cable.

Can anyone shed some light?

--
Dave Friend
Not speaking for IBM
dave DOT friend AT home DOT com

2. printing/graphics question

3. Checking modem rx and tx speeds during ppp

4. ftp TO slip-connected linux

5. MSN messenger file tx/rx problem via a Linux gateway?

6. need source for xflip

7. 3c905 100Mbps: RX-ERR RX-DRP RX-OVR?

8. need help with bridging

9. non-PLIP parallel file transfer?

10. Corruption of files during FTP transfer?

11. sz hangs during file transfer

12. High latency PPP link (Unresponsive) during file transfer Question

13. networkdevices RX/TX status