> On Aug 7, 2025, at 3:17 AM, rain_alex via lists.fd.io 
> <wangdaobei=gmail....@lists.fd.io> wrote:
> 
> It can be confirmed that what nginx sends is the last ACK packet from the 
> previous stream.  I don't think there is a problem with VPP because it 
> accesses the same source port every 9 seconds. You can also see this in my 
> attachment.

Not sure there’s any problem, the only one that came to mind was the potential 
of rejected port reuse in time-wait by linux. If linux did not transition to 
time-wait because of dropped packets and lost retransmits and vpp timed out in 
close-wait/last-ack, then the problem is different. Those timeouts are 
configurable, in case you want to explore debugging this. 

>  
> The only thing I doubt about the kernel program is that at times nginx resets 
> the fin ack packet of the kernel program. I am not an expert in host stack. 
> Can I request the other party to send a reset instead of an ack by changing 
> or adding something in the fin ack packet?

No, normally the peer stack should reset only if the state is wrong and/or the 
peer application was unable to receive all data. So for instance, if nginx 
closed the connection and had unread data, that would explain the reset. A full 
close from nginx followed by more data from vpp, e.g., ssl shutdown from 
client, might force a reset as well. 

>  
> My current solution is to immediately resend the syn packet instead of 
> waiting for RTO time when receiving an ack packet that does not meet the 
> requirements in the syn_sent state. If necessary, I can publish the code.

Not sure about the solution here yet, but can review code and help drive 
towards a solution. 

Regards,
Florin

>  
> Regards,
> Rain
>  
>  
> 
> 
> 

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#26250): https://lists.fd.io/g/vpp-dev/message/26250
Mute This Topic: https://lists.fd.io/mt/114562621/21656
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/leave/14379924/21656/631435203/xyzzy 
[arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to