On Tue, 8 Feb 2022, Christopher Morrow wrote:
you know what you COULD do though... probe it with DNS requests, and then you know, test the service being offered, and still know that 'the internet is not on fire'.
What?!? Use UDP to test the Internet? How would you even know if the Internet was fine but some router didn't like how your packet smelled and dropped it? ;-) Seriously though, if ICMP is becoming the problem this thread seems to believe, TCP rather than UDP is probably a better judge of the "availability of the Internet" as the remote end is going to attempt to respond. Though I cannot argue that lack of DNS also can indicate why Chicken Little is perturbed. I don't have any issues with ICMP generally, though I'm usually sending such packets to systems and servers and networks I control or have permission/access to. For people that don't have access to multiple servers dotted around the Internet, is it time for them to move away from ICMP and start using HTTP HEAD TCP requests to well-known websites to determine if a route is available and functioning? That's a lot more data when multiplied by a few million queries per second, just to check that the Internet is up... but also less likely to get filtered or throttled to the point where you get no response, even though the sky is not falling. Beckman --------------------------------------------------------------------------- Peter Beckman Internet Guy beckman@angryox.com https://www.angryox.com/ ---------------------------------------------------------------------------