Martin,
There are many dup ACKs #1 in the trace. There was no dup ACK #2 and #3, so it is kind of difficult to talk about "fast recovery" here, isn't it?
I just try to understand when seeing such "abnormality" from a customer's capture.
Reagrds,
PV
--- On Tue, 4/13/10, Martin Visser <martinvisser99@xxxxxxxxx> wrote:
From: Martin Visser <martinvisser99@xxxxxxxxx>
Subject: Re: [Wireshark-users] Dup ACK #1
To: "Community support list for Wireshark" <wireshark-users@xxxxxxxxxxxxx>
Date: Tuesday, April 13, 2010, 8:19 PM
Vincent,
It is likely to be a congestion recovery mechanism (TCP fast recovery). I don't see any Window size change. Are only some packets Dup ACKs?
A port-mirror or other capturing issue would show as all packets from one end being duplicated. Also if you open the IP header you should see unique IP identifier which would prove they are unique packets generated by the workstation.
You haven't indicated whether you actually a problem to solve, or whether it is just something of interest in the capture that you see.
Regards, Martin MartinVisser99@xxxxxxxxx
On Wed, Apr 14, 2010 at 2:52 AM, vincent paul <amoteluro@xxxxxxxxx> wrote:
Hi Martin,
How are you?
Please find attached some traces showing the DUP ACK # 1(immediately from user's side only). Could you please educate what is/are potential problem(s):
1) congestion ??? How can we explain these immediate ACKs from TCP protocol implementation/requirements?
2) Could it be "mirror port problem"?
But we captured traffic using wireshark on customers' workstations, already connected to their network. This meant they did not prepare a port span connection for us to run test and capture with wireshark.
I greatly appreciate your help and expertise.
Regards,
PV --- On Wed, 3/31/10, Martin Visser <martinvisser99@xxxxxxxxx> wrote:
From: Martin Visser <martinvisser99@xxxxxxxxx>
Subject: Re: [Wireshark-users] Dup ACK #1
To: "Community support list for Wireshark" <wireshark-users@xxxxxxxxxxxxx>
Date: Wednesday, March 31, 2010, 6:40 PM
Vincent,
You indicated that you are trying to "understand any problem when a customer using our application across Internet.". You need to describe more clearly what the problem you are seeing is for us to be able to help. Is it unreliable (not connecting/disconnecting) or slower throughput than expected, or some other error reported by the application or user.
From what you have described it seems like you probably have normal congestion going on, and TCP is using one of the various mechanisms it has to recover from it. Without seeing a packet capture (probably you should be this at client and server end if possible) as well as an understanding of the network topology (is it LAN/WAN or Internet?) it is pretty hard to diagnose.
Regards, Martin MartinVisser99@xxxxxxxxx
On Wed, Mar 31, 2010 at 11:15 PM, vincent paul <amoteluro@xxxxxxxxx> wrote:
|
Hi Martin,
I google around I found one people had the same problem(similar to mine at server side) with wireshark capture, but there was no resolution. Please take a look and give some thought.
Duplicate ACKs are pretty common when you have congested network. If the client is expecting more data and doesn't receive it in time 9maybe 250ms) then it will send another ACK in the hope that the data it is missing/waiting on will be resent.
You haven't indicated what problem you are chasing. If you send a small sample capture that would be useful.
Regards, Martin MartinVisser99@xxxxxxxxx
On Wed, Mar 31, 2010 at 3:58 PM, vincent paul <amoteluro@xxxxxxxxx> wrote:
|
Hi All ,
I looked at two traces captured at user's side: one going thru proxy and one bypassing proxy and observed a lot of Dup ACK #1. Both traces are traffic of HTTP download's file from server. I have the following observations and could not find any explanation
1)Traffic going thru proxy: User always sent double ACKs (one ACK(len=0) and its Dup ACK #1(len=0) immediately). No Dup ACK problem from server side
2)Traffic bypassing proxy: server, most of the time, sent out double ACK (ACK(len>0) and its Dup ACK #1 (len=0)) with a time period. This means:
Server---> user: seq=1000 Ack=210 len= 1460
Server----> user (dup Ack #1) seq=2460, Ack =210 len=0
.
.
.
Normal TCP data transfer from server for a while(kind of frequency) , then Server sends out double Acks again.
But there were also some time intervals, the data transfer from server looked normal without any double Acks from server
In this case, no Dup Ack problem from user's side.
I appreciate your help very much.
regards,
PV
|
___________________________________________________________________________
Sent via: Wireshark-users mailing list <wireshark-users@xxxxxxxxxxxxx>
Archives: http://www.wireshark.org/lists/wireshark-users
Unsubscribe: https://wireshark.org/mailman/options/wireshark-users
mailto:wireshark-users-request@xxxxxxxxxxxxx?subject=unsubscribe
-----Inline
Attachment Follows-----
|
___________________________________________________________________________
Sent via: Wireshark-users mailing list <wireshark-users@xxxxxxxxxxxxx>
Archives: http://www.wireshark.org/lists/wireshark-users
Unsubscribe: https://wireshark.org/mailman/options/wireshark-users
mailto:wireshark-users-request@xxxxxxxxxxxxx?subject=unsubscribe
-----Inline Attachment Follows-----
|
___________________________________________________________________________
Sent via: Wireshark-users mailing list <wireshark-users@xxxxxxxxxxxxx>
Archives: http://www.wireshark.org/lists/wireshark-users
Unsubscribe: https://wireshark.org/mailman/options/wireshark-users
mailto:wireshark-users-request@xxxxxxxxxxxxx?subject=unsubscribe
-----Inline Attachment Follows-----
|