by Phillip

New About Yours API Help
1.4 KB, Plain text
Hi guys,

As a long-time Linux sysadmin, I decided to spin up a FreeBSD VM to join to the public NTP Pool as an opportunity to start playing with FreeBSD (outside of FreeNAS).

Obviously having a publicly accessible NTP server creates a lot of network traffic; perhaps too much from what I'm seeing.  Forgive me if I'm telling you what you already know, but public servers in the NTP pool are monitored for their availability, then given a corresponding score up to 20.  Only servers with a score above 10 are added to the pool.

The score for my host's availability is a saw-tooth pattern on the graph (attached -- blue line is the score), which suggests to me:
    1) My score increases to above 10; my host is added to the pool.
    2) Traffic increases to the server.
    3) Some kind of limit is being reached on my host.
    4) Monitoring host can't get responses; score goes back down under 10; my host is removed from pool and traffic drops off.
    5) Monitoring host can contact my host again; Go to step 1.

The fun part is this only happens with IPv4 traffic.  The same host via IPv6 has a score of 20 almost perfectly for months now (also attached).  Obviously there is much less IPv6 traffic compared to IPv4, which is what makes me think it's some kind of specific kernel or similar limit, as opposed to network/CPU resource limitations.

Do you have any suggestions where to be looking to debug this?

Cheers from AU/Melbourne,
~p
Pasted 7 months ago — Expires in 154 days
URL: http://dpaste.com/0ZSMVWH