[nos-bbs] Interested in REPEAT packet statistics?
Jay Nugent
jjn at nuge.com
Wed Apr 14 05:23:31 EDT 2010
Greetings,
On Wed, 14 Apr 2010, Kenneth C Coe Jr wrote:
> Sorry, Skip, I didn't realize you were waiting for an answer. I was
> still waiting for the rest of the question. Actually, I still am.
>
> As usual for your posts, half the information that is needed. You simply
> cannot boil a complex problem down to one or two sentences and expect an
> answer worth its weight in salt.
>
> The reason I snapped at you (trite or not), is simple. I have been
> watching emails come from you for a while, pointing out what you say are
> problems with the way the network is designed. You keep posting these
> snippets of "data" and saying you are requesting an answer to your
> question, making a habit of ignoring other peoples requests for the
> right information and just recycling the same old mess of numbers for
> your next question (or proposal, or whatever it is).
>
> I have seen a lot of good information proposed to you in past emails,
> and you appeared to have ignored it or not even heard. You certainly
> didn't hear mine, because you replied to my answer about only posting a
> small piece of the information with another snippet. Did you really
> expect an answer, or are you just trying to get yelled at?
>
> As far as the dumps, just post them somewhere and tell us where they
> are. We can download them...
>
> Ken
I will add to what both Ken and Barry have stated -- the information
you provide is only partial. I will restate that we need the
tools/commands used in your data collection (NOT some half-assed home-brew
script that collects only *portions* of the trace information) and we need
a clear starting point - what your settings are in AUTOEXEC.NOS, etc.
No matter whether your path is VC or DATAGRAM, the *LINK* layer needs
to be clean. As Barry pointed out, your T1 timer appears to be too short.
And your own comments confirm this - you see one packet go out, no ACK
received, more "RETRIES" (these are NOT "duplicates"), and then finally 3
ACKS come back at you for the same DATA frame. By this I would say your
timer settings are TOO short and the SRTT is trying to re-adjust to the
actual path round trip time once it learns how long it really takes for
the DATA/ACK exchange.
Back to the VC -vs- DG issue. Either way, the AX.25 "Link Layer" MUST
be working correctly or you will see problems. You should always start
diagnostics from the *BOTTOM* of the OSI Seven Layer model.
Layer 1 - Physical
------------------
We asked what your RF path looked like (you replied full quieting), we
asked if your radios were deviating correctly (you replied that both ends
were set for 3.5 KHz IDC). The only thing not confirmed it that the
transmitters and receivers are all within 100 Hz of Freq-center and that
your modem tones are properly calibrated. Also that there is no 60Hz hum
on the signal (bad filter caps can nail ya every time).
Layer 2 - Link
--------------
This is where AX.25 L2V2 protocol comes into play. This layer MUST
work perfectly for the upper layers (IP & Network) to pass data without
error or retry. So start us off by showing how good or bad plain AX.25
packets flow across this path. Don't convolute the problem with
additional higher layer traffic.
Only once your Link Layer is clean can we begin to troubleshoot Layer 3
- Packet layer (IP) issues...
So please stop side-stepping good engineering diagnostics and get on
with it! Else more and more people will simply begin to ignore your
posts for what they are --- a Phishing expedition...
Thank You,
--- Jay Nugent WB8TKL
> > >> What does the fact you receive dup ACK tell you about your
> > >> own node?
> > >>
> > > That your own node has a T1/Retry timer set to expire at least
> > > three times faster than it should be?
Train how you will Operate, and you will Operate how you were Trained.
+------------------------------------------------------------------------+
| Jay Nugent jjn at nuge.com (734)484-5105 (734)649-0850/Cell |
| Nugent Telecommunications [www.nuge.com] |
| Internet Consulting/Linux SysAdmin/Engineering & Design/ISP Reseller |
| ISP Monitoring [www.ispmonitor.org] ISP & Modem Performance Monitoring |
| Web-Pegasus [www.webpegasus.com] Web Hosting/DNS Hosting/Shell Accts|
+------------------------------------------------------------------------+
5:01am up 22 days, 2:32, 4 users, load average: 1.16, 0.40, 0.26
More information about the nos-bbs
mailing list