Open matteoraggi opened 7 years ago
My interpretation of what the original author wrote in README:
The granularity of HTTP header "date" is 1 second. So even in an ideal environment, the max possible error of a single request is 0.5 seconds.
Burst mode tries to send N requests uniformly in 1 second. Based on how many of them return T, and how many return T+1, it can estimate the millisecond part, by computing the average.
I didn't use it, but ran against 4 different hosts instead. Similar idea about accuracy, but better connectivity and stability are more important in my use case.
Thanks to clarify! So the only way to grow the time quality is the -b option. I can't figure how ntpd could better of htpdate...
I'm afraid that your conclusions are totally the opposite of what I wrote :(
Aha, I am very often back to firewall that I don't want to loose time configure about, so I would like to use htpdate instead of ntpd. Which is the option to use multiple hosts for htpdate? Is it possible to use both multiple hosts and -b option too? I cna't use NTP because of the firewalls locking it, right?
Read README. Yes. Ask the owner of "the firewalls", or try by yourself.
nono, I can't manage the firewalls for every situation, I prefer to use htpdate instead of ntpd. But why ntpd is more precise of htpdate?
Smaller granularity. Please read my 2nd comment.
You are right, but NTP is locked from firewalls, thanks for the patience, i foudn another and better solution maybe: RadioClock! I just need to find a hardware and not a DIY and a software for debian/raspbian.
The accuracy of htpdate is at least -+0.5 seconds and better with multiple servers, but huw muc hbetter? the only way to improve the accuracy is this option? -b Burst mode uses multiple polls for each web server to enhance accuracy.