time.gov reply from NIST
Sounds like there was a change made to the time.gov code recently that could have caused the error we noticed. I replied to Andrew that I didn’t remember the time.gov site loading any slower or faster than usual.
Jerry,
Thanks for writing. I have had a few reports of incorrect time, but they were from people comparing to cell phones and WWVB clocks. I have not been able to replicate any errors, but I had our network folks verify that the servers serving the web content are synchronizing with NTPd to NIST servers. I trust time-nuts implicitly - if you guys say there's a problem, I believe it.
It would be difficult to believe that the server clocks could be on that far, and then corrected, and then off again. If a refresh fixes it, then I'm confident that it is a network problem. The app corrects for half the round-trip delay, assuming that to be a good estimate for the one way delay from the web server. However, if there is a network bottleneck or vary slow packet transfer in one direction, then the correction will not be accurate. Have you happened to notice if it looked like it took several seconds to connect after making the web request?
I might have to take out the correction and just report the delay, how it used to be. This is also how the widget operates: http://time.gov/widget/widget.html http://time.gov/widget/widget.html So a check of the widget at the time the HTML5 application is showing an error I would expect it to be correct.
Thanks for the information! Please keep in touch.
Andrew Novick
NIST
(xxx) xxx - xxxx phone removed by Jerry
On 5/12/2017 11:15 AM, Jerry Hancock wrote:
Hello,
I am a member of a group of individuals called time-nuts. Most of us have atomic clocks or at a minimum, GPS disciplined oscillators that record and display time with a variance below 10 nanoseconds. Yesterday around 14:00 Pacific Time I noticed while setting a new watch that time.gov http://time.gov/ was as much as 5 seconds slow. After reporting this on the time-nuts site, other members found the same discrepancy. This was corrected later in the day. At least one member of our group reported that refreshing the site corrected the error. I found that by running the flash application that also corrected the error. Since around 18:00 Pacific yesterday the error has been corrected.
Can you account for the discrepancy?
Thank you,
Jerry Hancock
Thanks for this post, Jerry. Could you please forward my post to Andrew?
As you can see below, I have some issues about the widget I want to
discuss with him, and I don't have his email address. I visit NIST
Boulder every couple of months for business, but I'm never meeting with
people in his group.
I noticed unusual errors (of up to about 1 second) when the thread first
appeared here, and also noticed different errors when using the NIST
widget on my personal website. But tonight (as I'm posting), I'm
listening to 5 MHz WWV from the Dallas TX area and see no visible error
(within my human ability) comparing listening on an AM receiver to WWV
while watching these websites:
(1) http://nist.time.gov/
(2) http://time.gov/widget/widget.html <html5 version>
(3) http://time.gov/widget/ <requires Flash>
All three of these URL's work on most desktop browsers. The first two
don't need Flash, but (3) uses Flash. So (3) won't work on a mobile
Apple device which doesn't support Flash (such as an iOS device such as
an iPad, iPod, or iPhone).
In August 2011 (which I believe was shortly after the (3) Flash widget
was first published), I found a bug in the (3) Flash widget which
prevented the display between 12:00:00 AM and 12:59:50 AM (or 00:00:00
and 00:59:00 in 24-hr mode). The display instead showed PM (in 12-hr
mode) during that interval. I have an email I sent myself about this bug
but I can't find any email traffic to or from NIST at that time. It was
fixed long ago, so they may have fixed it without my complaint or I
might have spoken to them on the telephone.
I find that the Flash widget (3) works fine when the code is placed on
my website, but instructions are not available to do this for the
non-Flash (2) widget. I wish NIST would update the (2) non-Flash widget
so that it could be easily installed on another website. If anyone knows
how to do this, please let me know! I'm hoping Andrew can assist me.
Bill Byrom N5BB
----- Original message -----
From: Jerry Hancock jerry@hanler.com
To: Discussion of precise time and frequency measurement
time-nuts@febo.com
Subject: [time-nuts] time.gov reply from NIST
Date: Fri, 12 May 2017 17:45:45 -0700
Sounds like there was a change made to the time.gov code recently that
could have caused the error we noticed. I replied to Andrew that I
didn’t remember the time.gov site loading any slower or faster than
usual.
Jerry,
Thanks for writing. I have had a few reports of incorrect time, but
they were from people comparing to cell phones and WWVB clocks. I have
not been able to replicate any errors, but I had our network folks
verify that the servers serving the web content are synchronizing with
NTPd to NIST servers. I trust time-nuts implicitly - if you guys say
there's a problem, I believe it.
It would be difficult to believe that the server clocks could be on that
far, and then corrected, and then off again. If a refresh fixes it,
then I'm confident that it is a network problem. The app corrects for
half the round-trip delay, assuming that to be a good estimate for the
one way delay from the web server. However, if there is a network
bottleneck or vary slow packet transfer in one direction, then the
correction will not be accurate. Have you happened to notice if it
looked like it took several seconds to connect after making the web
request?
I might have to take out the correction and just report the delay, how
it used to be. This is also how the widget operates:
http://time.gov/widget/widget.html http://time.gov/widget/widget.html
So a check of the widget at the time the HTML5 application is showing an
error I would expect it to be correct.
Thanks for the information! Please keep in touch.
Andrew Novick
NIST
(xxx) xxx - xxxx phone removed by Jerry
On 5/12/2017 11:15 AM, Jerry Hancock wrote:
Hello,
I am a member of a group of individuals called time-nuts. Most of us have atomic clocks or at a minimum, GPS disciplined oscillators that record and display time with a variance below 10 nanoseconds. Yesterday around 14:00 Pacific Time I noticed while setting a new watch that time.gov http://time.gov/ was as much as 5 seconds slow. After reporting this on the time-nuts site, other members found the same discrepancy. This was corrected later in the day. At least one member of our group reported that refreshing the site corrected the error. I found that by running the flash application that also corrected the error. Since around 18:00 Pacific yesterday the error has been corrected.
Can you account for the discrepancy?
Thank you,
Jerry Hancock
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.