ARP confirmed timestamp update on TCP data flow vs keep-alive

ARP confirmed timestamp update on TCP data flow vs keep-alive

[Steffen Heil (Mailinglisten)]

Hi


I have a question regarding confirm timestamps in the ARP table.
If this is the wrong place, please tell me.

For established TCP connection with data flowing, the ARP table is updated, such that (for some time) no further ARP requests are required for later data packets. (Observed on a live system, and read here [1].)

This seems to work only for payload data, not for TCP keep-alive.
Is there a way to update the confirmed timestamp as well, if TCP keep-alive packets are received?

I already have a TCP keep-alive of 30s for thousands of connections, but now I also have ARP requests.
I know that I could modify base_reachable_time_ms (which I am already doing), but there is a trade-off of not detecting stale entries versus sending too many requests. If the keep-alive packets would keep the ARP entry alive, I could keep a low value for base_reachable_time_ms.


Best regards,
  Steffen



[1] https://stackoverflow.com/questions/15372011/configuring-arp-age-timeout/15511117#15511117
"the kernel will sometimes change timeout values based on positive feedback from higher level protocols"