Closed Worty closed 4 years ago
Hi @Worty,
I've noticed the issue myself, and was pondering a similar solution in order to make dns resolution stronger.
I managed to avoid the blocking host
call by setting the mtr rounds variable to 10 as opposed to the initial 5, which gives me back hostnames in every test case i ran. It looks like that on your system perhaps DNS resolution takes a little longer.
Before merging your PR, I'd like to know if you get any results by increasing that variable to higher values, so as to give mtr (and your nameserver, I guess) more time to resolve.
Could you please let me know?
Thanks for your contribution!
Well its very weird im running a local dnsmasq for dns caching so i usally get 0-2 ms response time but even with mtr -l -c 50 www.github.com
im not getting any rDNS answers.
Tested on Debian this time instead of Archlinux:
By using host
for rDNS you could reduce mtr_rounds back to 5 (or even lower on my tests) for a lower total runtime
I've implemented your suggestions (dropped dnslines in favor of host
lookups + reduced mtr rounds), alongside more speed improvements (mtr now runs asynchronously inside a coprocess), and speed gains are really good in my tests. Please let me know if everything works well for you. Thanks a lot for your feedback!
The rDNS lookup wasnt working for me because mtr doesnt return the hostnames in -l mode (because of latency i guess?). With this patch its working for me. Please tell me what you think about it.
Without patch:
With patch:
![07-08-2020_21 06 58_8e147127feebd4ef680b88c2b28716c9](https://user-images.githubusercontent.com/6840978/89680042-7da6f380-d8f2-11ea-983f-545eced7d11f.png)
btw this is a really great tool