Several people are trying to use either SQM or my script to solve their latency issues on their networks. To help in debugging those issues, I've written a simple Julia script which will traceroute to a given site, then ping all the sites on that traceroute for a given time, and plot the latency as a time-series. Tested only on Linux running standard linux traceroute and ping tools. Though there's no reason it shouldn't work in say WSL on Windows.
For those of you who are having mysterious latency issues such as @segal_72 you might want to try it out to debug some of those issues. This is VERY rough and not guaranteed to work at all, so I'd appreciate some testing and feedback. I'm very happy to add a bunch of features to this as time goes on, it's complimentary to the gaming script also in the same github repo.
To use it, you'll need to download Julia 1.7-rc1 at https://julialang.org/downloads/ and then run julia at the command line, and have it load the script. The first time you run, you'll want to uncomment the first few lines which makes julia grab the needed packages from the julia package server.
cd into the directory where you've grabbed my file, then run julia at the command line... then at the julia> prompt type as follows:
julia> include("juliapingplot.jl")
The FIRST time you run it it will take a while to grab the needed packages... After that, you can comment out those first lines, and it'll take a shorter (but still nontrivial) time to just load the packages... Julia is a just in time compiled language and has an issue with "time to first plot" so be a bit patient it may be some seconds before the plot screen appears.
You can adjust which site it pings via editing the script... I'm going to bed but you guys who are in a different part of the globe may try it out and give some basic feedback.
Yepp, you really mean that.... I did try with the current stable (1.6) to no avail (I want to keep installing stuff like julia via homebrew on my mac*)
That reminds me that "ability to read" often is insufficient if not exercised.
*) I assume that apple's ping, with its lack of support for D is not going to work, but I did not even get that far....
Hmm. Actually the only redundant part is the PKG.add, once you've done that you don't need to keep doing it. But keep the using Pkg and the activate line uncommented.
I'll put some logic in to make commenting and uncommenting irrelevant. The bigger issue is it runs and does nothing with the uncommented version I think perhaps because the Mac traceroute or ping are not working right for this script
as you can see even at 800Mbps or so of speedtest, my latency increase remains down in the 5ms range except for one outlier from one site along the way (which may well be that site responding less quickly rather than anything else). The code outputs a data frame with all the measurements and could be made to write that to a CSV file for later analysis. It can also potentially do statistical analysis of that data to determine where in the chain the delays occur, what the average delay increase is... etc. Those are the kinds of things I'd like to add ultimately. In particular for some of what @segal_72 has been experiencing, I suspect that he's got congestion along the path rather than at the local connection. Collecting this kind of data can ultimately help to understand what's going wrong.
In order to make it work, the script does need to parse the output of both "traceroute" and "ping"... my traceroute outputs like this:
traceroute to www.google.com (142.250.188.228), 30 hops max, 60 byte packets
1 10.79.1.6 0.268 ms 0.234 ms 0.308 ms
2 192.168.1.254 2.987 ms 1.265 ms 4.953 ms
3 162.207.92.1 5.319 ms 11.280 ms 5.690 ms
4 70.232.229.68 6.284 ms 12.283 ms 5.920 ms
5 12.242.115.21 12.660 ms 11.913 ms 13.252 ms
6 12.255.10.176 11.570 ms 17.743 ms 12.827 ms
7 * * 10.252.217.126 15.254 ms
8 142.251.60.129 13.688 ms 13.564 ms 13.573 ms
9 142.250.188.228 13.981 ms 142.251.60.129 13.357 ms 142.250.188.228 13.037 ms
and my ping like this:
PING www.google.com(2607:f8b0:4007:80a::2004) 56 data bytes
[1633974125.855457] 64 bytes from 2607:f8b0:4007:80a::2004: icmp_seq=1 ttl=116 time=4.71 ms
[1633974126.856720] 64 bytes from 2607:f8b0:4007:80a::2004: icmp_seq=2 ttl=116 time=4.19 ms
[1633974127.858339] 64 bytes from 2607:f8b0:4007:80a::2004: icmp_seq=3 ttl=116 time=4.54 ms
[1633974128.859661] 64 bytes from 2607:f8b0:4007:80a::2004: icmp_seq=4 ttl=116 time=4.27 ms
[1633974129.862127] 64 bytes from 2607:f8b0:4007:80a::2004: icmp_seq=5 ttl=116 time=5.37 ms
[1633974130.862500] 64 bytes from 2607:f8b0:4007:80a::2004: icmp_seq=6 ttl=116 time=4.28 ms
[1633974131.863939] 64 bytes from 2607:f8b0:4007:80a::2004: icmp_seq=7 ttl=116 time=4.36 ms
[1633974132.866574] 64 bytes from 2607:f8b0:4007:80a::2004: icmp_seq=8 ttl=116 time=5.07 ms
[1633974133.866879] 64 bytes from 2607:f8b0:4007:80a::2004: icmp_seq=9 ttl=116 time=4.24 ms
[1633974134.868568] 64 bytes from 2607:f8b0:4007:80a::2004: icmp_seq=10 ttl=116 time=4.64 ms
--- www.google.com ping statistics ---
10 packets transmitted, 10 received, 0% packet loss, time 9013ms
rtt min/avg/max/mdev = 4.192/4.567/5.367/0.370 ms
So those are the formats it's expecting. If someone has issues, please send the format of your traceroute and ping output so we can deduce how to modify the script to read that data properly.
user@work-horse:~/.julia$ traceroute -n -w 0.5 google.com
traceroute to google.com (142.250.181.206), 30 hops max, 60 byte packets
1 192.168.42.1 0.283 ms 0.264 ms 0.255 ms
2 62.52.200.147 9.308 ms 9.488 ms 9.821 ms
3 62.53.12.14 9.764 ms 62.53.12.12 9.738 ms 9.178 ms
4 62.53.25.59 9.075 ms 9.050 ms 9.023 ms
5 74.125.48.102 9.283 ms 9.055 ms 9.265 ms
6 * * *
7 209.85.251.206 10.693 ms 209.85.251.130 9.652 ms 209.85.251.206 10.636 ms
8 209.85.240.161 9.590 ms 108.170.253.85 9.971 ms 209.85.240.83 9.446 ms
9 108.170.253.49 10.461 ms 142.250.181.206 9.433 ms 9.403 ms
similar to yours, except hop 6 where mine only has 3 entries, while yours has four...
user@work-horse:~/CODE/julia/julia-1.7.0-rc1/bin$ ping -n -D -w 10 www.google.com
PING www.google.com(2a00:1450:4005:802::2004) 56 data bytes
[1633977962.957324] 64 bytes from 2a00:1450:4005:802::2004: icmp_seq=1 ttl=118 time=9.64 ms
[1633977963.958362] 64 bytes from 2a00:1450:4005:802::2004: icmp_seq=2 ttl=118 time=8.91 ms
[1633977964.959678] 64 bytes from 2a00:1450:4005:802::2004: icmp_seq=3 ttl=118 time=9.19 ms
[1633977965.960913] 64 bytes from 2a00:1450:4005:802::2004: icmp_seq=4 ttl=118 time=9.10 ms
[1633977966.962372] 64 bytes from 2a00:1450:4005:802::2004: icmp_seq=5 ttl=118 time=9.37 ms
[1633977967.963654] 64 bytes from 2a00:1450:4005:802::2004: icmp_seq=6 ttl=118 time=9.17 ms
[1633977968.964961] 64 bytes from 2a00:1450:4005:802::2004: icmp_seq=7 ttl=118 time=9.16 ms
[1633977969.966174] 64 bytes from 2a00:1450:4005:802::2004: icmp_seq=8 ttl=118 time=9.09 ms
[1633977970.967463] 64 bytes from 2a00:1450:4005:802::2004: icmp_seq=9 ttl=118 time=9.16 ms
[1633977971.968717] 64 bytes from 2a00:1450:4005:802::2004: icmp_seq=10 ttl=118 time=9.12 ms
--- www.google.com ping statistics ---
10 packets transmitted, 10 received, 0% packet loss, time 9012ms
rtt min/avg/max/mdev = 8.908/9.190/9.639/0.184 ms
@@ -22,6 +22,9 @@ function gethops(dname)
while true
if(!eof(trout))
line = lstrip(readline(trout));
+ if match(r"[0-9.]+ ms",line) === nothing
+ continue
+ end
s = split(line,r"[ *]+")
push!(hops,(n=s[1],addr=s[2],ms=s[3]))
#@show hops
Check to see if the file is corrupted or something. There is no "<" character on line 7 of the script. In fact there is no "<" character anywhere in the file.
I did try with the current stable (1.6) to no avail 