[pskmail] Re: Server stomping on client.
- From: Steve <stevez@xxxxxxxxxx>
- To: pskmail@xxxxxxxxxxxxx
- Date: Tue, 08 May 2012 20:52:24 +1000
Thanks John, you're helping me understand this.
The haveSOH:0 I assume means that the server doesn't think it saw a SOH.
OK, some timing things I've observed. 100% repeatable is the server
transmitting two packets in quick succession, just after a connect. The
second one always clobbers a client transmission.
After a connect, and the server telling me the server version and how
many emails await, the client will reply.
The server will then begin, what I assume is the keep-alive two and fro.
In the example just run, this happened.
The server sent <us><soh>1$s! B026<eot>
This triggered the client to transmit, but only a second or two into the
client transmission...
the server sent <us><soh>1$s!!!E0E6<eot> - right over the top of the
client.
Interestingly, in this case FLDigi didn't show a <soh> from the client
in between the short server transmissions (it does sometimes though).
Maybe further increase TX Delay on the server?
I feel I'm getting somewhere thanks to your help.
Steve.
On 05/08/2012 07:25 PM, John Douyere wrote:
Steve,
The server is the master therefore only the server can talk over the
client from a timeout perspective.
So only two possibilities:
1. The client responds before the end of the server's TX. That is
possible with some modes like MFSK modes where the decoding can happen
before the end of TX. Solution for that is the increase in client
TXDelay but 2 seconds is enough in my opinion.
2. The server does not wait until the client's response. Since the
client is slave it will wait indefinitely until the server tells it to
respond. I suspect that is the case here since you say they talk over
each other.
The only reasons I know of which would create that is if
a) the server has not received the frame all together from the client
and times out or b) it has missed the SOH as reported by your log
extract and it timesout or c)the client is way to slow to respond. The
delay between the client's EOT showing on the screen and the TX should
not be more than the client's TX delay plus about 1 or 2 seconds, so
max 4 seconds in your case.
It could be also a combination of b) and c). Give us some approximate
timing values between EOT as seen on the receiving side and start of TX.
Also, what modes have you tried? Can you restrict the mode list to
PSK500R only and try.
73, John
On Tue, May 8, 2012 at 6:50 PM, Steve <stevez@xxxxxxxxxx
<mailto:stevez@xxxxxxxxxx>> wrote:
Hello,
The server stomping all over the client is 100% reproducible in my
case. The packet below is the entry in the console (that
pskmail_server is run from) that coincides with a transmission
that _ALWAYS_ transmits over the top of the client. It's just
after I connect and the server sends a message telling the client
that "it cannot find any mail".
After that first collision the whole session becomes a mess with
both sides talking over each other, seeing the resultant frame
loss, and so drop down into Thor22 mode. They still talk over the
top of each other though.
0 Sending poll with havesoh: 0, maxidle: 11.998258103927,
Shortidle: 8.95053974787394, currentidle: 9
Other related posts:
