jeff@heurikon.UUCP (Jeffrey Mattox) (09/13/85)
Sometimes, due to a hardware timing problem, we loose a character
while receiving data. The message handshaking stops and uucico
at our end times out after 10 seconds. We send a short "blip"
(a retransmit request, I think) and sometimes the remote starts
transmission again. However, sometimes the remote *ignores* the
retransmit request and then we give up after 5 tries.
NORMAL: Almost continuous data coming from the remote system,
with ACK/NAK two or three times per second. (1200 baud)
ABNORMAL: We stop sending ACKs. About one second later the
remote host stops sending us data. Then there is no
activity for almost exactly 10 seconds. At that point
our system sends a short burst and the remote begins to
send data again.
REALLY ABNORMAL: We don't get any response from the remote and
we finally give up and disconnect.
This is the output of uucico -x9. There are a number of "alarm"
lines where the 10 second timeout occurs
rec h->cntl 37777777603
send 40
rec h->cntl 37777777613
send 41
rec h->cntl 37777777623
send 42
rec h->cntl 37777777633
send 43
rec h->cntl 171
bad header 31,h->ccntl 40
rec h->cntl 37777777653
rec h->cntl 37777777663
alarm 6
send 43
alarm 7
send 43
alarm 8
send 43
alarm 9
send 43
alarm 10
send 43
tries = 5
send 10
send 10
cntrl - -1
enter us_sst, status is : 10
s.sysname : uwvax
send OO -1,imsg >\ 20 OOOOOO\ 0 exit code 0
I realize that we are at fault in the first place for missing
a character, but it seems to me that the remote has a bug, too,
because it cannot be woken up. Anybody seen this before?
We are a System V, talking (usually) to BSD 4.2 (uwvax).
--
/"""\ Jeffrey Mattox, Heurikon Corp, Madison, WI
|O.O| {harpo, hao, philabs}!seismo!uwvax!heurikon!jeff (news & mail)
\_=_/ ihnp4!heurikon!jeff (mail - best)