Do You Have The Time?

RetroGeek writes: "This ZDNet article talks about the perils of the PC clock. And (something I did not know) that Windows XP and Mac OS X both automatically get a time stamp from MicroSoft and Apple respectively. At any rate, my home firewall gets the time every hour from the NIST servers, then each of the machines on my LAN query the time server daemon on the firewall. That way all my home network machines have the same time. And latency on the LAN is next to zero. Now if I can only get my VCR connected. Anyone else running a time server?" So how do you get the time?

Comments

How do I get the time?

[Ethelred Unraed]

I look at a clock. Or maybe my (wind-up) wristwatch.

Sheesh. Geeks. If it ain't digital, it ain't.

Cheers,

Ethelred

Re:How do I get the time?

[Bagheera]

I look at the display on my dash, wrist, cell, or whatever happens to be available in line of sight. For my organic needs, "accurate within a few minutes" is accurate enough.

For my LAN, I have two machines running ntpd, getting their sync from two different sets of time servers. The other machines on the LAN sync to the two local time servers.

My digital needs require better than "within a few minutes" accuracy.

Re:How do I get the time?

[neuroticia]

I look briefly at the time when I set up my computer, ensure that it's within the proper range of "It's light out" or "It's dark out" and leave it be.

Late for a meeting? "Oh my god! The clock on my PC was wrong! Damned XP time-synch decided I was in Hungary.

Seriously, though, I prefer setting the time myself from my watch or from the microwave in the lounge, or from calling out to a co-worker "Hey Sam! Got the time?"

I just feel odd about letting *anything* remote change any setting on my computer, even if it's just the time. ESPECIALLY if I'm on Windows. I mean... How long before there's a clock-virus? :p

-Sara

Microsecond accuracy for $25

[shoppa]

1. Roofmounted Trimble SVeeSix-CM3 GPS receiver with microsecond-accurate pulse-per-second output: $24.95.
2. Network Time Protocol synchronization software: Free

Re:Microsecond accuracy for $25

[ceejayoz]

3. Having the correct time, always: Priceless.

Re:Microsecond accuracy for $25

[mizhi]

That's probably alot better than the roofmounted sun-dial and array of lightsensors that I have.

Re:Microsecond accuracy for $25

[shoppa]

Did you just have to build a serial cable for that, or did you have to build some other interface electronics to make it talk to your computer?

The GPS output signals are CMOS/TTL level, not RS-232, so I put them through an MAX 232converter before they come out of the box on the roof and run downstairs to the PC. This is not exotic stuff; TTL to RS-232 converters are pretty much 30 year old technology.

The protocol is just plain async serial, so no special electronics to encode/decode.

Re:Microsecond accuracy for $25

[Above]

I see I'm not the only one, so details!

I've been looking for a cheap GPS time receiver that interfaces with standard ntpd nicely. I can see you're on to something with these referb units, and the serial converter you mention. I could probably figure this out given enough time, but a few more details of what you bought, and how you hooked it up would be really nice. If I can get this going for

IP spoofing target

[RichMan]

Ok, this is bad. A somewhat critical state of the OS is dependant on a blindly connected service. Please tell me the time server is authenticated fully and unbreakably. Hah.
Just wait for
1) MS to implement expirable licenses on all software
2) someone to break the authentication service
3) IP spoofing of the time server to a clock set 100 years in the future when everyones time based license has expired

The result is instant crippling of all MS licenses!

Re:IP spoofing target

[FearUncertaintyDoubt]

You forgot:

4) ???
5) Profit

Re:IP spoofing target

[reemul]

You have to manually edit the registry to use any servers other than the two defaults - one MS, and one US gov't. Not too tricky, but there is no interface and no documentation to add servers, so most everyone will end up using the defaults. The part that annoys me the most is that the update schedule isn't configurable, it always just checks one week after the last check. You can do a manual update, but then the next update will be, you guessed it, another full week later.

Re:IP spoofing target

[Monkeyman334]

I don't think so :) If you didn't want your subscription to expire you could set your computer to look at a local ntp server instead of MS, and have your subscription never expire. I'm guessing they do a little more than if (date expire_date) { on the computer to see if your license expired. And don't make bogus claims you can't back up, quit with the "all they have to do is..." attitude. You post is so subjective it sounds like you don't have the slightest clue what it takes to accomplish what you're saying.

Re:IP spoofing target

[MaxVlast]

Sheesh. On Mac OS X, you can simply type in a new server. Editing the registry? That's a futuristic, easy-to-use OS, isn't it?

the benefits of accurate timekeeping

[lakeland]

Personally I arbitarialy declare the firewall as having the same time and use cron to update everyone from that. Since latency between machines is almost equal, everybody is out by the same amount.

Before anybody thinks it is silly to keep clocks tightly synchronised, try running NFS without it and you'll run into no end of problems. Even as little as one second will cause errors with make. The key is that all clocks must read the same, not that they need to be correct.

Oh, and don't get fooled into thinking you can accurately synchronise against those atomic clocks. The algorithms they use to average results make a number of incorrect assumptions that will result in you being out by a small constant amount, about as much as if you'd synchronised off an ordinary clock.

Re:the benefits of accurate timekeeping

[swright]

Since latency between machines is almost equal, everybody is out by the same amount.

NTP uses a nifty little algorithm to compensate for network latency. I forget the details but it makes little difference whether you sync from a LAN box or from one on the other side of the planet.

School server time

[agent oranje]

At my school, a time server is set up to keep the computers on the network within a certain range of time. I believe the purpose of this is for security, as we can't renew our kerberos tickets if our time is more than X minutes from the server's specified time.

around the clock

[jean-guy69]

So how do you get the time?

using one of these ?

Time for my VCR

[the eric conspiracy]

Some VCRs including my JVC can get a time signal that is broadcasted by PBS stations via cable. It's wonderful to never have to set that puppy.Combined with ntp for my computers, and WWV for my stand alone clocks (so called 'atomic alarm clocks' I am down to one clock that I have to set - my wristwatch.

Re:Time for my VCR

[Mr. Sketch]

I am down to one clock that I have to set - my wristwatch.

Not if you had one of these.

Re:Time for my VCR

[Technician]

Check your local mass merchant. On the West Coast we have Fred Meyer. They carry a line in the $50 range. The casio brand is high enough volume to provide economy of volume while providing a quality product. They have the usual utility watch features including water resistant.
A list of the watches can be seen here in the $50 price range. It includes LaCross and Casio brands.
http://www.gadgets4sure.com/cgi-bin/ePage s.filerea der?3d24dd6b001506400000c0a8013a0569+EN/catalogs/1 23581

However if you are looking for pure geek appeal, check out the solar powered ceramic case model for over a grand here..
http://www.shoplifestyle.com/store/product .asp?dep t%5Fid=26%5Fid=4158=0VK0N50BE5XM8GVTV55B2GRBB1W2DP RC

Re:Time for my VCR

[MrCreosote]

'My watch is accurate to 1 second in 1 million years'

'So what time is it now?'

'Uhh, about quarter past.'

Re:Time for my VCR

[11thangel]

Yes, those are nice, but when you wear one, you have to watch out for those nasty Grabites.

--End Bad Sci-fi movie reference--

Re:Time for my VCR

[Zeinfeld]

Would you really spend $300 dollars on a watch?

Ever wondered why so many consultants wear $5000 stainless steel and gold Rolexes? Its because it is a way to indicate to the customer that you don't work for cheap and the $3K per day or whatever you are charging is the going rate.

So no I would not wear a $300 dollar watch, its too expensive to wear to the beach and I certainly could not afford to wear something that cheap to work.

In Windows? NetTime...

[krez]

There's a nice open-source utility at Sourceforge (http://sourceforge.net/projects/nettime/) that I use at work on my Windows machine.

I like it because it's simple, unobtrusive, and invisible once it's installed.

Re:In Windows? NetTime...

[rehannan]

In the interest of multiple opinions, I like Automachron. It's a lot smaller than NetTime (130KB vs. 2MB) and is also unobtrusive. You can make it invisible or stick it in your system tray.

Here's a screenshot of it running on my system.

Simple

[Mr. Sketch]

time.org.

I can even get the date too :)

Re:Simple

[thilmony]

but can you get A date?

Re:Simple

[Mr. Sketch]

Isn't that part of the date?

Actually from that site you can view calendars for various years. Only in the future though, no past calendars unfortunately.

Re:Simple

[Phork]

are you really that bad at math. is subtracting 8(or 7) that hard?

Re:Simple

[Phork]

no, i didn't have to guess, i said 8(or 7) because PST is -8, but PDT is -7, right now the offset is -7.

Re:Simple

[MiTEG]

Maybe time.gov is better..... ;) check this out

Does it worry anyone else that it's the Texas flag on the time.gov site?

http://www.ntp.org

[MavEtJu]

Go to http://www.ntp.org to get all your time-synchronisation questions answered.

Also for in- or near-Germany living people: http://www.dcf77.de. Wish I knew it was a german-specific service before I came to .au and found out that my DCF77 receiver didn't work here...

VCR Timers

[BlueFall]

Some newer VCRs (and therefore probably DVD players) have this feature that allows them to set their time to a time signal on a certain channel, usually public television in the US. The station transmits the time via XDS (extended data services). Maybe you could set up something with a TV card on your time server...

oh dear... too... much... hacking... ;-)

Re:VCR Timers

[Phork]

PBS's uasualy handles the time broadcast in most US markets, and it is notoriously innaccurate.

Re:ntpdate [server] in crontab...

[Dr. Ion]

Consider running a proper NTP daemon instead.

It has the advantage of not jerking your clock around every time you sync. It makes calculated "smooth" adjustments to keep your clock accurate. It can also use multiple servers.

It's the difference between a perfectly-ticking clock, and one that gets manually reset twice a day to make it (temporarily) accurate.

The biggest impact this will have is if you do file access across the network or need your timestamps to be reliable. Depending on how much your clock drifts, that ntpdate adjustment could back up several seconds. This can wreck havok on timestamp-dependent things, like "make".

Most ntpd distributions make this easier to set up than a crontab entry anyway. :)

The intelligent way

[halftrack]

Look at my left wrist, and see if it checks with my pc-clock. If not I set it - that is; the pc-clock.

Maintaining a medium-size net of clocks

[angio]

As part of the Resilient Overlay Networks project at MIT, I maintain a testbed of about 20 nodes, most of which have GPS-based time synchronization. We've started using a really fun little box from EndRun Technologies called the Praecis Ct. It gets GPS time that's being rebroadcast by cellular CDMA base stations. They provide accuracy to about 10 microseconds, and don't require a roof antenna -- anywhere you can get CDMA cellular service, you can use these things. They're kind of pricey (about $1k), but they're completely easy to use and set up. For more general information about NTP and things, see ntp.org, which mtaintains a nice FAQ about things-ntp.

For a few of the european hosts, we use GPS time receivers, primarily the Motorolla Oncore UT+ kits. You can get eval units of these, google around. They're nearly as easy to use, but do require a kernel config change.

It's really kind of addictive playing with time. :-) And you get spoiled by never having any clock weirdness on any of your machines...

I found...

[IanBevan]

..that the Microsoft time server was 3 minutes slow ! This was about 2 weeks ago. I checked it against both another time server, and then the UK speaking clock (dial 123 in the UK) which is synchronised with Greenwich. As a result, I disabled the time synch (right click on the time in the system tray, Adjust Date Time, Internet tab, uncheck the box). I now use the time synchronisation feature that comes with the Dynip client.
Since the MS time synch is enabled by default, they really should make sure their server farm has the correct time :(

Re:I found...

[DNAGuy]

It is important to note, that your XP box will actually synch with your domain controller(s) if it is part of a domain. If the time is out by more than 5 minutes (by default), your Kerberos tickets can be expired before you get them.

ntpd and k9

[fava]

I run ntpd on my firewall machine and set it to broadcast over the local network once a minute.

The local machines run a small (64k) utility called K9 which listens for the broadcast and sets the time accordingly. I found most time clients for windows were very large and much to bloated for what I wanted to do. K9 works perfectly. There is even source code available for your favorite flavour of *NIX

I have the time

[billsf]

If you use UNIX, just set up ntpd. You are often
requested to inform the providers of stratum one
servers that you use them. Since most NTP
servers discriminate against end-user DSL and
cablemodem services, i offer a "stratum 2" service
for these people.

All told, all my friends have the time to a few
milli-seconds, a vast improvement over what the
local telco can offer.

As for Windoze, i know nothing, but believe
NTPD is somewhat functional.Time is very
important for UNIX and all secure services.

Re:I have the time

[Dwonis]

At work, we use Automachron on our Windows boxes to sync up to a local time server.

Coursey is a whinner...

[burnsy]

UNFORTUNATELY, the clients in Windows and Mac OS aren't ideal. They share two problems: First, they may not synchronize often enough.

That Coursey sure is a whinner and clearly he does little research. I took me 15 seconnds to find this at Google.

To control the number of seconds to wait between attempts to synchronize the system clock to an time source on the Internet using the following Windows XP...

Hive: HKEY_LOCAL_MACHINE
Key: SYSTEM\CurrentControlSet\Services\W32Time\TimeProv iders\NtpClient
Name: SpecialPollInterval
Type: REG_DWORD
Value: #secondsdesired default

Re:Coursey is a whinner...

[ceejayoz]

However, I suspect the time servers limit the number of requests from the same IP in a certain period of time, to avoid DoS attacks. So the point may still be valid...

Most cablemodem/DSL head-end routers have the time

[Dr. Ion]

If you're on cable or DSL, most of the upstream routers run proper NTP servers, and they're just a hop away. The bandwidth for running an NTP client is minimal.

To find the nearest NTP server, to a traceroute to a few non-local hosts. Then start at your nearest router and ping each one for a time server using something like 'ntptrace'.

Near-perfect accuracy, just a trickle of data, and your provider will thank you for using nearby machinery.

I use NTP like this.....

[Dr. Awktagon]

On your Red Hat Linux server/firewall/whatever (easily adapted to any NTP setup, really):

ntp.conf:
server time.apple.com
server tick.usno.navy.mil
server tock.usno.navy.mil
# In case the network is down
server 127.127.1.0
fudge 127.127.1.0 stratum 10

broadcastdelay 0.008
authenticate no

driftfile /etc/ntp/drift
pidfile /var/run/ntpd.pid
logfile /var/log/ntpd

and /etc/ntp/step-tickers has the IP addresses for those hosts, all one line (the Red Hat startup script uses these to set the clock at boot, in case it's WAY out of sync.):

17.254.0.27 192.5.41.40 192.5.41.41 17.254.0.26 17.254.0.31

Then on your LAN, have all your other machines use this machine as the time server. That's it! Never set a clock again.

It's important to have accurate time for many protocols, including HTTP, and also to timestamp your logs accurately for forensics and evidence.

For even more accurate and secure local timeservers, run a GPS antenna to your roof and buy one of these products.

Re:I use NTP like this.....

[synx]

interestingly enough Mac OS X also uses NTP.

it is unix after all....

Re:I use NTP like this.....

[Phork]

almost all time synchronization is done with ntp.

Re:I use NTP like this.....

[thogard]

Don't use the master clocks unless you need that kind of accuracy (then ask them and they will give the the names of better servers).

Do a traceroute out out of your net and see if you can find a few other servers.
$ ntpdate -v -u 64.39.2.65
will tell you if its running ntp or not. Pick a few of your upstream and go with that. If you have several upstream routes/providers then ntp will make sure you get the correct time if one of them gets way out of sync.

Re:I use NTP like this.....

[Eriq64]

>>It's important to have accurate time for many protocols, including HTTP, and also to timestamp your logs accurately for forensics and evidence.

For forensics & evidence?!!

So when they find all the bodies, you can prove to the millisecond that you were 20 miles away logged on to www.blahblah.blah, and couldn't have possibly bludgeoned all those mimes to death with a summer sausage?

ntpdate

[Combuchan]

nexus:~# ntpdate time.nist.gov
4 Jul 15:17:34 ntpdate[26989]: adjust time server 192.43.244.18 offset 0.000626 sec
nexus:~# date
Thu Jul 4 15:17:22 MST 2002


It's 3:17 PM right now. So yes, I know what time it is. Debian users can apt-get install ntp or ntpdate... it should be part of the base system in freebsd, and the NTP homepage is http://www.eecis.udel.edu/~ntp/

Sh

rdate

[BrookHarty]

Ya, With multiple PC's in the house, my windows boxs always had the correct time(sa.windows.com i think is the time server). On my unix boxes I just use rdate, "rdate -s time-nw.nist.gov" and everything is set. Was thinking about setting up a ntp server, but it would use time-nw.nist.gov also, might as well cut that step out.
-
Verizon uses thin copper on city streets... = no dsl.

Re:Is this an XP thing?

[ceejayoz]

afaik it's only in XP... hasn't been in any Win2000 or WinME installations I've played with, but it's definitely in every WinXP installation I've seen.

clock setting

[frovingslosh]

Pre-Internet I used a program that would make a quick long distance call to the naval observatory over the modem to set the clock. When I first got Internet access I started using a program called Atomic Clock, but over the years fewer and fewer time servers seemed to support the protocol it used, and eventually the very few I could find were obviously in need of some attention to their clocks themselves, they were drastically off. I'm currently using a little program for the PC called NTPC. I've had to occasionally track down a new time server when the predefined ones became unavailable, but otherwise it works fine.

Can the M$ time sync for XP be disabled, or is this just another way for them to impose Bill's vision on us all?

Re:clock setting

[Reziac]

[laughing] Man, all this hoo-rah about time -- okay, if you're running a server farm I can see the point, but for a home box?? If my personal computers are within 5-10 minutes of the correct time (or each other) that's good enough. Nothing in my life needs fraction-of-a-second accuracy, and I'm not anal enough to worry over it either. :)

Comment removed

[account_deleted]

Comment removed based on user account deletion

Comment removed

[account_deleted]

Comment removed based on user account deletion

a potential problem

[CAIMLAS]

I know several people that set their clocks ahead of time so as to make themselves think that they're running late, when in actuality, they're on time. If MS or Apple changes this time to the 'correct' time, this could cause people to actually -be- late. Imagine the dilema: you come in late, and lose your job. Is that MS's or Apple's fault for changing your time on you w/o your permission? Or your's for using their OS?

Re:a potential problem

[stripes]

Is that MS's or Apple's fault for changing your time on you w/o your permission?

Well in Apple's case at least it might be your fault for not going to "Date and Time" panel and either unchecking "Use a network time server", or pointing at a NTP server that keeps your kinda time (yes OSX uses real NTP, and yes, they let you choose any NTP server you like).

Or much better...for not changing the timezone files so you live 7 hours and 50 minutes ahead of GMT not 8 hours...

Re:a potential problem

[CAIMLAS]

For me, maybe. I'm talking about the type of people that's mentality requires the clock be set 10 minutes fast or such - they're also generally the type that don't know what their computer does, in my experience.

Re:a potential problem

[Buck2]

My SO is like this and it drives me fucking nuts having all the clocks set anywhere from 5 to 20 minutes in advance. She has it all set up so that she "gains time" from when she gets out of bed to when she gets in her car, ie. the bedroom clock is 20 minutes ahead, the kitchen 15, and the car clock 10.

It's not worth a serious argument though because I don't usually use clocks anyway. It's just annoying as hell when one of my buds asks me the time and I'm like, "Uh, well that clock is 10 minutes ahead, I think. Er, wait, that one's fifteen, it's the other one that's ten. Well, it's like 3:15 plus or minus 10 plus or minus 5."

Re:a potential problem

[bnenning]

In Mac OS X, you have to turn automatic NTP synchronization off before you can manually adjust the time. At least when using the Date & Time preference pane; you can also run "date" from a command line but in that case you should know what you're doing.

Re:a potential problem

[CAIMLAS]

Hrm. And I thought I had it bad. My mom sets all the clocks in the house 20 minutes fast.

I think the only 'irritating' time-related traits my fiancee has (which I also happen to share, I might add) are 1) having 2 alarm clocks for the morning, and repeatedly resetting them over a period of time until awaking (say, half an hour, an hour... there-about), and 2) waiting to the very last possible moment to leave for an engagement, and then running in a couple minutes late (to no dire effect)

Re:a potential problem

[CAIMLAS]

There are certain reasons that I would never find myself in your bedroom, let alone necessitating an alarm clock, that I believe my obtainment of a girlfriend leaves obvious. :P

my setup

[deviator]

Netware 5.1 server gets time from several NTP servers (i.e. tick.usno.navy.mil, tock.usno.navy.mil, etc.) and triangulates "correct" time from averaging out the sources. (Netware actually has the most intricate and cool time synchronization system built-in because NDS depends heavily on accurate timestamps)

Windows-based workstations automatically set clock to time on Netware server using Novell-supplied file client software (Client32) when they login.

Linux boxes get time from Netware server using NTP.

MacOSX laptop gets time from Apple using NTP (it's mobile & physically travels to many different networks. :)

btw, Microsoft has no concept of time synchronization. Throwing an NTP client into Win2K & WinXP isn't exactly what I'd call "enterprise-class time synchronization." I've struggled for years using a variety of techniques to keep clocks accurate on mid-sized Windows-based networks. Novell by _default_ synchronizes the local PC clock with the main login server. You actually have to override this feature if you want to do it yourself. It saves so much effort...

Re:ntpdate [server] in crontab...

[rodgerd]

Actually, that's not always too polite. The public servers are all pretty much heavily loaded (even down to strata 2), so I hope you're syncing off a time server on your ISP.

VCR

[AntiNorm]

My VCR is able to auto-set its clock based on XDS data that is sent along with closed caption information.

Re:Aint that just the way...

[rodgerd]

Even many of these are way too heavily loaded. Many ISPs run ntpd on some of their servers; point at them, instead.

Re:ntpdate [server] in crontab...

[renehollan]

I second this. I have a 24x7 DSL connection on a firewalled and NATted static IP address, and connect to a not-too-far NTP server (the owner of which is nice enough to let anyone connect for only the asking of permission).

That takes care of one PC, and therefore all PCs.

My VCR gets time from a broadcast stream, and my satellite receiver from the satellite. (I always thought satellite receivers should have built-in NTP servers and ethernet ports for, among other things, program guides, but I digress).

Now, the microwave and oven clocks, as well as my alarm clock, are dumb in this regard.

Re:Aint that just the way...

[TeddyR]

They dont even need to use srtatum 2 servers if they are on cable, or have a responsible isp.

Many isp's have an ntp server that they use for their own equipment. Ask them what they use.

Most Cisco routers with IOS 11.3 or higher (methinks...) can act as ntp servers for an end node.

Most cable providers "head end" equipment are also NTP servers. (Part of the DOCSIS standard requires that the cablemodems sync their clocks when they get their config files).

Most Linux/RH users can traceroute to somewhere... and then use ntptrace on each hop that traceroute shows to see if the device is an ntpserver.

Use the closest one that has the correct time.. [because unfortunately, some ISPs dont know how to properly/fully configure their equipment.]

...s XP and Mac OS X both automatica...

[MavEtJu]

MacOS 8.6.x already had this feature.

The benefit of ntpd

[Paranoid]

The benefit with running a server like ntpd is that its always running. It'll run a query once every 5 minutes or so, and rather than just simply resetting your clock, it'll statistically derive the real time accounting for network lag and randomness and all of that. It'll reliably get better-than-a-millisecond accuracy, given a few days of this. Plus, it can use the same mechanism to detect drifts and inaccuracies in your PC's own clock, and tweak the kernel to compensate. Plus, if you have a pool of peer machines, they can help eachother detect drift more quickly (without having to resort to longterm statistical analysis).

For these reasons, I run ntpd on most of my machines, rather than some set-and-exit cron job.

Synch interval change

[happypizzaguy]

"The default--and unchangeable-- synchronization interval for Windows XP is one week."

This isn't entirely true: while there is no way to change the synch setting in Windows using the UI, but a simple change of a number in the Registry will give the desired results:

To change the interval that Windows updates the time using the internet time servers via regedit, navigate to:

1. HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services \W32Time\TimeProviders\NtpClient
2. Select "SpecialPollInterval"
3. Change decimal value from 604800 to a different value in seconds. i.e.: 172800 (2 Days) or 86400 (1 Day) and so on.

apt-get install chrony

[John Hasler]

The default configuration works with a dialup.

Coincidence

[Large Green Mallard]

Just yesterday at work I was talking with a researcher about this.. he was showing me an NTP server he made, using two DGPS units and some embedded ethernet controllers.. he said the accuracy on it was about 40 nanoseconds from UTC..

That should probably be suitable I think :)

http://truetime.net sells some rack GPS-based NTP Servers too.. but I don' know the price.

AutoClock

[RAVasquez]

My older Mac runs AutoClock, which computes the mean deviation between logins to NIST, then automatically adjusts the clock whenever it drifts out of sync. Neat.

Whoa there!

[piecewise]

"The people who use NIST's Internet Time Service range from Windows XP and Mac System X users"

If they're not informed enough to call it Mac OS X (and the last time apple used "System" was before Mac OS 8 came along... which was a longggggggggg time ago), then I don't trust the rest of the article, either. So I don't care, and I'm going to use Network Time to set my clock right now, just to spite that stupid Microsoft Windowed TP-user.

NTP here, too

[LinuxHam]

My firewall config had a Linux box on the outside with a crossover cable into a Linksys. The perimeter firewall has now come inside. I run NTP on 6 Linux boxes and used to run w32time on an NT Server.

Only problem is, I could never get the firewall (or any single NTP client) to become a server for the rest of the network, so my boxes are not very polite right now. I've read tons of docs and google hits, and they all seem to indicate that once an NTP client syncs, it will also become a server. Not so here.

What, if anything needs to be done to make a sync'ed NTP client also act as a server? Thanks..

How do I update?

[xanadu-xtroot.com]

Easy:

[root@aragorn /]# cat /etc/cron.hourly/timesync
rdate -s ns.coop.net&


Just search google for "Stratum 1" servers and look for your timezone in the list.

ntp.nasa.gov

[foniksonik]

I use NASA's ntp server. ntp.nasa.gov

OS X Date/Time has an option to use ntp and you can set which server. ntp.apple.com is the default but I prefer to use a server where time is very important for everything they do. NASA seems to fit the bill.

Other ntp servers like any military server are probably similarly effective. I wouldn't necessarily trust a commercially owned and run ntp server though.

Re:ntp.nasa.gov

[rtaylor]

Yes, I can see how that tenth of a second can make a huge difference. I could be late for a meeting by that much -- it wouldn't be good.

NTP all the way

[gregbaker]

My firewall runs ntpd to sync its time with one of the public time servers in Canada. All of my Unix-ish machines run ntpd to synchronize with that; Windows machines run Tardis on startup to sync.

A trick to find nearby time servers (other than looking at a list): run ntptrace on a nearby, well-administered Unix machine. Find the last machine that's inside the organization--that will be the one they sync with the outside world. Run ntpq on that machine and type peers. You'll see a list of the NTP servers that it queries. Put some of those in your /etc/ntp.conf and you're good.

Don't Do That

[John Hasler]

"...my home firewall gets the time every hour from the NIST servers,..."

Don't use stratum one servers for your home network. It's wasteful and unnecessary. Use a stratum 2 or higher server or your ISP's server.

Re:Don't Do That

[stor]

Indeed.

Please listen to this guy. Stratum 2 or higher is perfectly fine for a corporate network let alone a home network.

Please don't overload the stratum 1 servers, it really doesn't buy you anything.

As others have suggested, please set up one machine to run an ntp server e.g. xntpd, and sync off the remote ntp server and have other hosts sync off the local ntp server. This is desirable from your POV anyway. Why send NTP traffic across the internet if you don't really have to? That means more traffic (yeah, negligable) and higher and unpredictable latency to the ntp server. That latency matters: ntp tries to compensate for it but of course the lower and more constant the latency the better.

Also if you're running linux (and perhaps other *nixes), learn about /sbin/hwclock --systohc. Very useful, especially as a cron job on the local ntp server.

Cheers
Andy

Re:Don't Do That

[leighklotz]

Except that ISPs rarely provide time servers and the stratum 2 servers all have "Don't use us without permission" disclaimers. The delegation system is broken; there's no incentive to provide new stratum 2 servers and no incentive for end users not to use stratum 1 since they're readily accessible and well-known. Try reading the UDEL howto; it gives a list of stratum 2 servers and almost all of them have dire warnings about not using them without prior notification.

Re:Don't Do That

[Phork]

They dont say prior permission, they prior notification, and they are many that dont say that. They server ops just want to know whose using there servers. If they really didnt want people using there servers they wouldnt be on the list, or they would be mareked as private, or they would be firewalled.

Re:Don't Do That

[AndroidCat]

Imagine if a significant percentage of XP users shift to NIST. They all sync at the same time every four hours, *thunk*! (I think it was 4 hours, I'm too lazy to go back and check.)

A random (done once and kept as a constant) offset within the four hours for each machine would be more polite.

I normally just set my computers off my old "digital" flip-leaf clock (based on power-line 60 Hz) The computer was fast by 16 seconds when I checked, which is close enough for my needs.

Re:Is this an XP thing?

[netsharc]

net help time (in W2K Command Prompt). The only GUI option is the option to start the service on bootup, it's under Administrative Tools, Services. You can even set the NNTP server it should connect to, I use my uni's server.

Time Server_s_, plural...

[The Monster]

The summary says that you have to get your sync from MS. Fortunately, this is untrue. I see two choices when I'm running XP:

• time.windows.com
• time.nist.gov

Take a wild guess which one I chose...

But if you want more choices than that:
This article inspired me to do some dumpster-diving in the Registry... Import this key/value:

[HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\ W32Time\TimeProviders\NtpServer]
"InputProvider"=dword:00000001

This allowed me to set my own choice of NTP server, and then synced from it. Like many other MS 'features', the
default can be changed, if you know how...

Re:Time Server_s_, plural...

[The Monster]

You're right. For reasons I can't explain, after changing the Registry back, it did indeed perform as you say. But I swear I tried this before, and it didn't work.

Re:Time Server_s_, plural...

[Emvious]

Why not just overtype the Server in the control panel?

Re:Time Server_s_, plural...

[ImaLamer]

There is always the option to not let it update.... that is usually the most secure.

The thing that bothers me is that my software firewall never asked to access the net to update the time. Of course it's because the core OS parts are all or nothing when it comes to internet access....

AboutTime

[IanA]

Connecting to NIST using SNTP.
Resolved address for NIST (192.43.244.18).
Received time (ping 63 ms), error -1 ms.
New time: Thursday, July 04, 2002 19:34:15.

AboutTime
(win32)

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:What about

[MaxVlast]

You can in Mac OS X. I can't fathom including such functionality that can't be turned off. What about the hapless PowerBook user who's machine doesn't shut down because he disconnected from the network while the NTP was trying to open a connection to the server?

Re:[SC0RE: -1, Microserf]

[AndrewHowe]

Excuse me, but you are ignorant. The operating system is almost irrelevant here. Any fool can have a clock accurate to a second or so (or more to my point, a lot more accurate... But most people will not quibble about a second). More accuracy is hard. NTP. Even with a decent protocol you have latency. It matters not what your operating system is. It matters more where on the planet your system resides. This is a known physical latency. There are relativistic effects... But seriously, don't worry about them... You will not notice... Trust me!

And you can hack the interval, too.

[The Monster]

A few more moments perusing the Registry reveals that you can also set the interval between syncs:

[HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\ W32Time\TimeProviders\NtpClient]
"SpecialPollInterval"=dword:00003840

The value '3840' there is hex for 14400, the number of seconds in 4 hours. Note that setting
the key won't affect the next, but the one after that will read this value to determine the time
for the one after that.

Best Windows 2000/98/etc utility - cmdtime

[rochlin]

After trying many time utilities for Windows which failed either in terms of simply getting the time sometimes or by being resource hogs (how hard can it be?), I found a great small footrprint freeware util called cmdtime.

It's highly configurable by batch file or commandline (it's a commandline tool in a windows world - which is much better for something you want to be unobtrusive) and just plain works.
cmdtime (and some other stuff)

Or, if you need something even better than NTP...

[jelson]

Wow, who would have thought that the topic of my PhD thesis would be on the front page?

Right now I'm doing research in very high precision time synchronization for very large numbers of very small things. My lab does work in sensor networks -- get a tiny little computer with a few sensors and a radio, sprinkle thousands of them out over a building or a battlefield or a forest. Have the network tell you where the fire started, where the enemy is lurking, which light bulb needs to be replaced, or a thousand other things.

You need very time sync to do lots of this stuff -- to track motion, for example. Our current testbed times the flight of sound to tell how far apart things are, and for that we need accuracy on the order of 10 microseconds between clocks.

My research right now centers around a new time sync scheme, called Reference Broadcast Synchronization, which in a recent study I showed is almost an order of magnitude more precise than NTP under the same conditions -- 5 microseconds between a group of nodes with a userspace implementation, and down to 1 microsecond in the in-kernel implementation (which is the resolution of the clock! I'll do better when I have a clock that ticks more than once a microsecond.)

NTP, even under "optimal" conditions -- very high query rate to a stratum 1 GPS-steered clock in our lab--- did no better than 50 microseconds. When we introduced high levels of congestion on the network, NTP degraded by a factor of 30 while RBS was almost unchanged.

Of course, NTP is still a fantastic protocol, and much better than trying to apply RBS to the Internet (which is basically impossible). But for tiny nodes that need very tight time sync, I say, we can do better :-).

Some recent papers you might like are here, including

• "Fine Grained Network Time Sync using Reference Broadcasts" -- the original RBS paper
• "Wireless Sensor Networks: A New Regime for time synchronization" -- my argument as to why NTP shouldn't be used for sensor networks
• "Locating nodes in time and space: A case study" -- description of our testbed that is capable of localizing objects down to 1cm by measuring time of flight of sound, combined with RBS time sync.

It's funny, I'm sitting in the lab right now, tinkering with the testbed when this article should come up.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:My XP box doesn't connect to Microsoft.

[zulux]

Windows stores settings in a nice fast indexed database. ...that frequently gets corrupted.

I understand the reasoning behind the 'windows registry' but all it does is mimics Unix text files in a filesystem hierarchy with 'keys' in a psuo-database hierarchy. /HT_SOME_UPERCASE_THING/sofware/windows/sytem/run /

is not any easier than /etc/inetd.conf ...plus Unix has the added benifit of comments in the text files, so you can document your changes and reasoning. As far as I know, you can't add comments to a Microsoft Windows registry.

As far as I know, it's not common to use tools to iterate over the windows registry. But it's standard practice to use common command line tols to change a particular configuration file over a network of Unix computers.

So for one computer and one operator, perhaps Windows REGEDIT so somehow easier, but propigating non-trivial changes over an entire network is easier in Unix.

Give Unix a try - there's a reason that it's the worlds most popluar network operating system.

Get lost?

[autocracy]

I mean, geeze. I know some people who aren't the best with directions, but you're the only one I've ever heard of that needs to know the Latitude and Longitude of his house to get back... Or is this just in case the foundation shifts?

Re:Get lost?

[thogard]

I've meet someone who uses GPS to find out how much his country (Sweeden) moves. He has a plot of most of Europes movments. The plots are on an sheet of paper in actual scale. Aparently Sweeden moves a few inches a year.

Re:What about

[ergo98]

NT 4 had time synchronization, but it came front and center with Windows 2000 (which automatically time sync with your domain controller, which itself should be configured to sync with a master authority) because of Kerberos as implemented in Windows 2000 and beyond : Time is one of the parameters, so if Bob's PC is 20 minutes ahead and his kerberos keys are by a different clock, security conflicts will arise (namely, his keys will be refused). Of course, anyone with XP can bring up the clock applet (double clicking on the clock in the system tray) and choose the Internet Time tab that allows them to change the server to a different one if they so desired (or disable it all together).

Two words

[grunby]

man rdate

- grunby

Atomic Wall Clock

[aztektum]

Radioshack has a wall clock that checks for (i forget the exact title) a radio signal that the Gov and NASA use to synchronize their time.

Click here!

noonch

Re:Atomic Wall Clock

[Phork]

It gets the signal from wwvb in colorado, which broadcasts the time on 60khz encoded in bcd using some odd modulation scheme. wwvb time is very accurate, at the trasmitter i think the accuracy is within a picosecond. If you know your location you can calculate the time it takes the signal to reach you from the transmitter, and get your time as accurate as the clock at the transmitter. WWVB is run by NIST. They also run two other radio stations, wwv and wwvh. WWV broadcasts from the same site in colorado that WWVB broadcasts from, but broadcasts a voice signal of the time, a pulse every second, as well as bcd and several other things. WWV broadacts on 2.5, 5, 10, 15, and 20 MHz. WWVH broadcasrs the same thing as WWV but is located in hawai, has a female voice instead of the male voice that WWV has, and doesnt broadcast on 20mhz. WWV rocks.

Re:My XP box doesn't connect to Microsoft.

[TummyX]

I don't know if one can add time servers (perhaps in the registry? never read anything about it), but it would be very nice to find out one could


Um, err... Just type the name of the timeserver in the textbox and it'll be added. No need to access the registry.

On Make

[os2fan]

Make only uses the timestamp to provide sequencing of dependancies. All "make" requires that all times come from the same clock, not that all clocks be accurate.

And since make is using file-stamps, this condition would be pretty much met by the clock that controls the fs where the data is stored. If there is more than one clock involved, then this becomes an issue.

The idea of synchronising clocks is that by being accurate, they will be "in step" with once another.

Someone compiling things on a computer where the standalone clock is used for both the fs and the system clocks.

Re:On Make

[Dr. Ion]

Imagine you're in the middle of running a make process when the local clock gets slammed with ntpdate (or rdate). If the clock was set back, it's quite possible that you now have recent files that are stamped slightly in the future. Perhaps a recently created file is even older than the parent files that generated them.

By using a one-shot clock adjustment like ntpdate, you're causing incongrous (not monotonic) timing. That can impact things that depend on one event being before another, like make.

I'm not making this up, since we actually had this problem on some old SunOS servers. If you run very long make processes (like X11 world) and you update the clocks regularly for sync, it's easy to get make all confused.
In any case, my point was that it's best to use more gradual sync programs, like ntpd, if you can.

Re:On Make

[os2fan]

The point of make is that it relies on a clock to form a sequence. Your point is perfectly correct, and I have also experienced it, too. But if the clock is not backtracking (eg slowed down), then make will be happy.

Make can also get slammed if more than one clock is involved: eg parallel compiling on different computers.

Re:[SC0RE: -1, Microserf]

[foobar104]

There's latency, and then there's relativity. When the server receives the time request, it takes the current timestamp and puts it in a network packet, which then trickles down the wire to the client. The client receives the packet and then knows what time it was at the server. That's latency. If you're NTP'ing over a dial-up connection from a distant server, the latency can be a second or more in worst cases. (NTP may have features to compensate for this; I couldn't say.)

Relativity affects the rate at which time runs for two observers in different inertial frames. It doesn't affect synchronization directly; if you ignore or compensate for latency, you can synchronize two clocks in different reference frames. But the clocks will start to drift apart immediately due to the different rates at which time passes in the two frames.

Now here's the cool thing. According to general relativity-- actually, according to my vague recollection of general relativity from a college semester more than ten years ago-- gravity affects the rate at which time passes in a reference frame. In other words, time runs more quickly in a high gravity field relative to a lower gravity field.

It's pretty well known that the local force of gravity varies measurably over the Earth's surface. Depending on where you are, the local force of gravity may be higher or lower.

So if you wanna get accurate, pick an NTP server in a region with a similar local G to yours.

HHOS. ;-)

Re:why go 3rd party ?

[TimeGuy]

Why go 3rd-party? Because, NET TIME (and Windows Time) is fraught with problems, particularly in a larger network/mixed environment. More detailed info about this at our site http://www.greyware.com/software/domaintime/produc t/w32time.asp Yes, I work for them, but I'd use this even if I didn't.

Re:My XP box doesn't connect to Microsoft.

[Russ Steffen]

Can you really trust running on the same computer to tell you when it's calling home? All these personal firewalls depend on an OS provided API to capture outgoing packets and alert you when something suspicious is happening. What if in the next service pack or "urgent security update" Microsoft decides that it's vitally important that Internet Explorer reports every single URL you visit. What stops them from replumbing the network stack so that Zone Alarm and TPF et al, never see that outgoing traffic. Nothing. You'd only be able to detect this with an external firewall, preferably one running on a trustworth platform, as the same dirty trick could be used to hide such "phone home" traffic from other machines as well.

Re:[SC0RE: -1, Microserf]

[AndrewHowe]

+5 Good advice ;) If you really care! LOL!

Simple time script

[bulletman]

Here's how I get and set the time:

date -u -s ` telnet 131.107.1.10 13 |grep UTC |cut -b16-23 `;hwclock --systohc

-- Stephen

Re:sync is one thing, but what about freq?

[John Hasler]

Once the clock is synchronized, why rely on a
relatively imprecise crystal to beat the drum?

Chrony measures the accuracy of the crystal and uses that information to "dead-reckon" when the Net connection is down. Thus the crystal only needs to be stable, not precise.

The 60hz AC line frequency is more accurate, or
so i've been told.

How accurate do you think the 60Hz from your UPS is?

rdate?

[zenyu]

Why not use ntpd?

Is it a security thing? There was a hole last year I think.

ntp tries to compensate for latency on the network, and you usually have two stratum 2 time servers on a decent sized network to triangulate between. At work we just use a single stratum 3 server (the main NFS server). But it uses two stratum 2 servers (tick and tock ;).

rdate with cron would work on Unix, but what about you're Mac OS 9 and Windows clients? Plus isn't it easier to just use the ntpd that's usually installed by default?

Re:Accurate timestamps

[damiam]

Off-topic: You're welcome to pledge whatever you want. That doesn't mean the rest of us should have to.

Adjtimex

[agentZ]

IIRC, This can be done on any system that system that supports adjtimex() . It's very handy, but not really new.

Re:In Windows? The Fourth Dimension

[wideangle]

I like Dimension4. Besides the Dr. Who reference, it's 112KB -- smaller than your Automachron.

Here's the screenshot.

Plus, it's coded by some guy at Microsoft. :

Will Any GPS Work

[suwain_2]

I happen to have a GPS sitting on my desk, that I haven't touched in a long time. (I put some batteries in it a few weeks ago to figure out my lat/longitude...)

What I'm wondering is... Will any GPS work, or is this is a specific feature of your GPS? I know that GPS uses very accurate timing to get its information, but I've never seen a way of getting it from a GPS. (I have a Garmin GPS 3 {Plus | Pro} here, a pretty common model.)

Although, now that I think of it, my GPS has one slot on the back, that can be used for either power or an external data link. If you ask me, it's a STUPID design -- the one time you'd actually want to use external power would be... when you had it connected to a computer. (I don't suppose there's a "hack" to run power and data into it at once?)

Re:Will Any GPS Work

[jrp2]

What I'm wondering is... Will any GPS work, or is this is a specific feature of your GPS?

It needs "NMEA" support. NMEA is just a standard for how to spit timestamp and time-pulses out the serial port in a consistent way. Many (most?) recent models do, some super-cheap and many older GPS models do not. You also need to get (or make) a cable to connect to your serial port and almost certainly will want to run this off a power supply and not batteries (and AC PSUs are generally not a normal accessory for a cheap GPS).

Re:Will Any GPS Work

[jrp2]

You need pps support which sends a pulse at the time the second actually changes.

Thanks for the added info. That is what I was referring to with "time pulses" in my post. Doesn't NMEA support assume "PPS" support, or is that an optional part of the spec. Just wondering as I assumed any GPS with NMEA support was OK for NTP.

Related: update daily

[coyote-san]

Another point is that it's unnecessary to update more often than daily except for the most exacting situations. Do you really need to keep your clocks synchronized to within milliseconds? I've found daily updates against a time server (which is sync'd to my ISP's NTP source) via a cron job running 'rdate' is good enough to keep my systems synced to within a second.

The other nice thing about this aproach is that it's easy to toss the Windows equivalence of 'rdate' into the startup scripts managed by Samba, so whenever a Windows box comes onto the network it's also synced.

Let xntpd do it's magic

[drix]

For Unix, don't fuss with ntpdate, cron, rdist, or anything use. Use xntp . Trust me on this one. You will gain a newfound appreciation of the complexities of network time synchronization after spending a good hour trying to figure out how to get this daemon to do the simplest of tasks, e.g. automatically keep your clock in sync. It's obvious from the docs and code that someone has spent way, way more time thinking about all this than you'll ever want to. XWindows, Linux, Sendmail, Apache, and xntp always come to mind when I think of free software projects that just freak me out because of how well done they are.

For Windows you can get ntp daemons but I find that, if something new appears in their system tray, users will fiddle with and break it every time. I use Samba and MS Windows networking built-in time sync, put in a startup script so it syncs on every boot. Clock drift on any modern computer is going to be negligible even if you're only syncing once every day or two.

Re:Let xntpd do it's magic

[Animats]

It's obvious from the docs and code that someone has spent way, way more time thinking about all this than you'll ever want to.

That's Dave Mills. He implemented TCP/IP long before Bill Joy, and designed several of the Internet protocols. But he always liked time synchronization.

He likes to watch what happens when a leap second is added. The power grid actually changes frequency to catch up. Every synchronous AC rotating machine on the grid has to make 60 extra turns. Mills has measured this; it takes about four hours for the grid to adjust.

A few years back, somebody at NIST forgot to throw the Daylight Savings Time switch, and the WWV signals didn't have the DST bit set for most of a day. Big embarassment.

On OSX...

[banky]

Just load netinfo manager and look for /config/ntp. From there you can use whatever server you want.

Re:On OSX...

[foniksonik]

Or just open up the old Date Time control panel and click on the "Network Time" tab and turn it on. Fill in ntp.nasa.gov and you're ready to go.

Re:My XP box doesn't connect to Microsoft.

[Zeinfeld]

In windows, once you know the setting's key/value you can easily change it with a few API calls. In Unix, you'd have to parse a text file insert/delete/change lines and then rewrite the file. Fun

You both missed the real problem with the UNIX approach - race conditions when two applications try to update the system configuration. This can really screw up a system because under UNIX the system config is in several files and you can be fairly certain the bozo who wrote at least one of the programs trying to update your config did not understand what locks are for.

I have seen many a UNIX system collapse in a gibbering heap due to corrupted config files.

I have ceased to be amazed at how people can praise inadequate engineering for its simplicity. Any problem can be given a simple solution if you ignore at least half of the problem.

The success of UNIX propagation says nothing for its utility. Herpes has also spread widely.

Re:Most cablemodem/DSL head-end routers have the t

[DragonWyatt]

Nice...

OT: WTF with no manpage for ntptrace (Redhat 7.2)? or info?

Check with your ISP

[Xtifr]

Your Internet Service Provider may well offer NTP as part of your service. They may not make a big deal of it, but it's probably available if you search their web pages or ask. I found a list of my ISP's NTP servers in their FAQ.

The advantage to using an officially provided host for NTP service is that it's less likely to disappear out from under you. One you hunt down on your own may be a temporary machine, or may be subject to change without notice.

but check your ISP first, eh?

[Xtifr]

Except that ISPs rarely provide time servers

They don't? I'll grant that I've never seen an ISP advertise their time server, but I've usually been able to either track one down by browsing their web pages, or, worst case, get the host info by emailing support. Of course, I do tend to use geek-friendly ISPs when I can, but I'd assume that most slashdotters would gravitate towards the geek-friendly ISPs. I'd strongly suggest you ask your ISP before assuming that they don't provide NTP servers. Chances are near 100% that your ISP uses NTP, at least internally.

Re:but check your ISP first, eh?

[Xtifr]

Um, what part of "you should ask" didn't you understand? Running ntptrace does not constitute asking -- in fact, I'd consider it slightly rude to do so without asking first.

"They don't advertise it" is not at all the same thing as "they don't provide it". And yes, it's certainly possible that they don't provide it, I suppose. Doesn't match my experience, but I've only had a couple of ISPs so far. However, it doesn't sound like you've ever asked.

SMTP and POP3 are things that customers demand, so of course they get advertised. Time services is something that probably only a handful of customers know about or care about, so it doesn't get advertised. But that doesn't mean it's not available if you ask!!

Re:but check your ISP first, eh?

[Xtifr]

I'm not actually complaining about what you did, I'm registering a very mild complaint about your advice, to wit, that you failed to mention that people should check with their ISPs first.

As for easy -- I did a quick search for "NTP" on Speakeasy's support database, and got an answer. Sending a quick email to support@ISP is pretty easy too. These are the easy approaches, and I think it's a bad idea to dismiss them without even trying. Don't assume your ISP doesn't offer NTP until you've checked, because it's very easy to check. If you've checked and the answer is "no", then that's fine, go probe for servers or whatever.

But the more people that ask for time services, the more likely that ISPs will offer time services as part of the package (like Speakeasy does), and the more likely that the answers will be on the FAQ list. (The name "Frequently Asked Questions" ought to be a clue here -- it only becomes a FAQ if it's asked. Frequently.) So asking is a good thing, even if you get negative results.

Bottom line, I don't object to what you did, I simply think that your advice is skipping what should be an obvious and easy first step.

Someone just tell me...

[Rombuu]

...and I've always wondered about this... why can I buy a digital watch for $10 that keeps better time than a Sun Starfire 15K? Or pretty much any other high end system for that matter? I mean, I would think for $1.5 Mill they could put a decent clock in the damn thing. For that matter, why does my desktop lose time so badly... I just don't get it.

Re:Someone just tell me...

[slim]

Maybe because Sun assume you'll be on a network, and if you care about the time you can run ntp...

Oddly enough...

[MbM]

My pc gets it's time set via PBS. Take a bt848 tv capture card capable of reading the "extended data services" line of video, bit of C and you can get a semi accurate time source.

VCR time

[plumby]

My VCR gets its time from the TV signal. Don't most videos have this now?

Easier way.

[justin_w_hall]

Well you COULD hack your registry, yes.

Or, you could just click into the field in your Internet Time dialog box (where it has time.windows.com and time.nist.gov) and type in the NTP server you'd like to use and click Update Now. Pointed it to my firewall and it worked fine. No sense in doing something the hard way. Especially if you're the pinnacle of laziness like me. :)

Microsecond accuracy?

[Paul Johnson]

Careful here. I once hooked up an NTP server to a GPS receiver (makers name omitted because I don't recall who made it) and found a 13 second error. Obviously the makers test suite hadn't included a check that the time it gave was right.

I note that on the page you are pointing to, the microsecond accuracy is described as "nominal". In other words, it isn't actually that accurate.

A purpse-made GPS receiver for time server applications is available, but costs rather more.

Paul.

Damn mobile phones...

[forgoil]

The GSM protocol can sync your mobile phone for you. No more setting the time, no more incorrect time. But do you think they actually put it there in all networks? No... We lack it here in Sweden, the country where GSM was invented. Pretty bad I'd say.

Re:Is this an XP thing?

[Sheridan]

NB. Microsoft hasn't exactly been RFC compliant in their NTP server/client interactions.

If you do a Google groups search for "NTP XP Mills" you'll find a host of articles detailing exactly what David L. Mills (Author of ntpd and the RFC1305) thinks of Microsoft's (intentionally?) b0rken implementation of NTP in WinXP this is one example

Re:Timestamping problem

[heikkile]

I think the traditional approach is to let the server measure the time used, like in the "old days". And let the server measure the latency, so that it knows how much to correct. Even this is not foolproof, as a malicious client could detect when the server is measuring the latency, and artificially boost that, so that the server would overcomensate, giving the player extra time. There is also the problem of unpredictable variations in the latency, which could be solved by measuring the latencies of every packet transmitted... Maybe a two-way system, where the server measures the latency to the client, and the client measures the latency to the server. The server recognizes these measurements, and adds a random extra latency, and if the client reports that wrong, it assumes there is foul play, and insists on using its own guesstimates of the timing. Hmm, trickier problem that I thought. Good luck!

Leap seconds, dude

[shoppa]

I once hooked up an NTP server to a GPS receiver (makers name omitted because I don't recall who made it) and found a 13 second error. Obviously the makers test suite hadn't included a check that the time it gave was right.

It was right. The GPS time epoch is 0000 UT on 6-Jan-1980. Since then UTC has had 13 leap seconds inserted. This offset is available in the NAV message; maybe the version of NTP you used was ignoring that message or maybe that particular GPS receiver didn't implement that message. (Actually, buggy firmware in GPS receivers has been a problem in the past.)

Re:What about

[Grab]

Solution for me (on Win98) is using the StayLive utility. It's a ticker to keep a dial-up link alive (to stop your telco cutting the link after 2 minutes of not doing anything), but it uses the various world clocks as its ticker, so it has the useful side-effect of keeping your PC clock perfect for as long as you're on the Net.

Re being reliant on MS, StayLive can connect to any of a dozen or so servers round the world (by default; more can be added if you know their IP addresses), MS is just one. So presumably XP could be set up the same way - there's bound to be an IP address settable somewhere. Oh, silly me, it's MS - what was I thinking? ;-)

Grab.

Re:Timestamping problem

[heikkile]

The server can't measure the latency all on it's own.

Of course not, measuring latency requires a packet to make a round trip. It does not have to be the same packet as the one used to tell the what move the user made. This all has to be built in the protocol.

Re:Most cablemodem/DSL head-end routers have the t

[thogard]

You ISP does this because they tend to run cisco gear and its easier to configure it to do NTP than to set the silly clock. If you care about your logs having the right times, its a good thing to have the clock set.

conf t
ntp master
ntp server 139.130.6.25
ntp server someother.server
(Please pick something better than my ISP's upstream router, this one won't be very good for you)

Re:Most cablemodem/DSL head-end routers have the t

[ajs]

Pretty cool. I've updated my /etc/ntp.conf and /etc/ntp/step-tickers accordingly. I used to just grab random entries from the official list of public NTP servers.

nano != 10^9

[tommck]

He said "Nano being 10 to the 9th power...".

That should be 10 to the -9th power!

T

GPS receiver

[Ricdude]

Garmin GPS receiver connected to the serial port, and a short script to ask it what time it is, and set system time accordingly. I'm not all that worried about the milliseconds of lag in actually setting the system time, but it doesn't require access to an ntp server, or even that my network be connected. Plus, once I programmed the data retrieval routines, what's programming one more unpacking routine for a data packet...

Re:[SC0RE: -1, Microserf]

[foobar104]

Oh, that's right. I stand corrected, and I feel silly for not remembering that. Thanks.

Re:You've got it backwards regarding gravity

[foobar104]

I already said this to an AC, but I feel dumb for getting that backwards. Thanks for correcting me.

My only excuse is that it was really late, and I was dazzled by the fireworks. ;-)

AtomTime95

[AnotherBlackHat]

I think there's a 98 version available now too.
Look here

Re:ntpdate [server] in crontab...

[dadragon]

Now, the microwave and oven clocks, as well as my alarm clock, are dumb in this regard.

I think I just found the reason why people want bluetooth in everything!

I originally thought it was stupid to have bluetooth in things like microwaves, but I've seen the light.

Re:Microsecond accuracy for free

[Isaac-Lew]

Doesn't NTP compensate for network latency?

Re:ntpdate [server] in crontab...

[renehollan]

I think I just found the reason why people want bluetooth in everything!

I came to this same conclusion.

Though, appliance-rich areas, like kitchens, and laundry rooms could do with a pre-wired network: either ethernet or HomePNA. HomePNA is probably easier for low bandwidth applications, like time sync, status checks, etc., and, if I'm not mistaken, can be piggybacked on POTS. Most people have POTS lines already pulled to the kitchen, the location of most dumb clocks. Heck, use a powerline based network, just get time to where it's displayed!

Java apps that report time

[Isaac-Lew]

Where can I find a good one? I've just set up a stratum 2 time server & it would be nice to have one in case someone actually goes to the URL (and I could put a link to www.ntp.org as well).

Re:Verizon CDMA time is inaccurate by 55 sec

[angio]

Interesting. Do you know which towers these are? I've encountered one Verizon tower that was off by 1 second, but I reported the problem, and they got it fixed shockingly quickly. All of my time collectors are happy right now, and I suspect some must be on Verizon (the one in New York is, for certain). The time zone is irrelevant; the raw signal sends out GPS time, which has no time zone.

As a safeguard against this, my boxes NTP peer with a subset of each other, and each box peers with at least one external, nearby stratum-1 timeserver. It's a fairly robust setup; overall, there are about 15 CDMA time receivers, 3 GPS receivers, and 13 external stratum-1 servers involved. We're susceptable to GPS problems because of the large GPS-derived presence in our network, though three of the sites do peer with NIST atomic clocks. But that's not too big a worry. No individual clock failure will hurt things much, except rendering the attached box useless.

Re:[SC0RE: -1, Microserf]

[geekoid]

so there's no time in space? cool, suddenly travelling from one galaxy to another got a lot faster.

Re:PC Clock anomaly

[alangmead]

The original IBM AT used a MC-146818A for its real time clock. Maxim/Dallas lists their DS12887A as a functionally equivalent chip. (with the addition of added non-volatile memory locations for added CMOS data. http://www.maxim-ic.com/quick_view2.cfm/qv_pk/2682 ) The seconds are accessed from address 00h on the chip. I guess its possible for some motherboards to use different chips, but maintain the same interface by presenting 0 for seconds, but I've never come across anything like that. Reading the time from the RTC is a somewhat convoluted process, so what most systems do is to read from the RTC once, and maintain an internal counter via a timer interrupt.

What you might be thinking of was MS-DOS's FAT filesystem, which stored file timestamps as the number of even seconds since Jan 1, 1980. Or maybe the fact the RTC only reports up to seconds (so being off by .5 seconds per boot.)

"changing the operating systems on your PC"

[JamieF]

>Working behind the scenes, a small government agency headquartered outside of Denver
>operates a network of 14 servers capable of changing the operating systems on your
>PC--and millions of others--in less than a second.

Sniff... sniff... what's that I smell? Ahh, the stench of yellow journalism.

You mean that NIST can change the OS on my G4 from MacOS X to LinuxPPC? On my PIII from Win2K to Win3.1? In less than a second? Amazing! They should sell this technology to Symantec for the next version of Norton Ghost! In my world it takes several minutes at best, maybe hours ("hours would seem like days" - Spock/TWoK), to install a new OS.

Oh, you mean, they can change the TIME VALUE STORED IN THE CLOCK. Is the clock part of the OS? No, as far as I can tell the clock is part of the HARDWARE. There is some clock functionality in the OS to get to the hardware clock and to provide time services to apps, but really, the clock is a hardware device. (I'm sure some /. kiddie will be happy to split hairs with me on that.) A more accurate lead paragraph would have ended:

>capable of changing the time on your PC--and millions of others--in less than a second.

Oh my god, no! NOOOOOOOOOO!

But wait a second. Doesn't he also say that WinXP syncs once a week, and OS X syncs at an unknown interval ('cos he didn't bother to research it and find out, he just looked at the UI and it doesn't say)? That means it really should say:

>capable of changing the time on your PC or Mac--and millions of others--
>in no more than a week. Unless you configure it otherwise, in which case your clock will be wrong.

OK, let's try that again: Oh my god, no! NOOOOOOOOOO!

Re:GPS based NNTP servers????

[alangmead]

I didn't think that the GPS satellites had the bandwidth to run a news feed.

One time Hayawatha Bray, the Boston Globe's technology columnist wrote an article about time syncing and he made the same mistake of saying NNTP when he meant NTP. Syncing time through Usenet has then become a running office joke. Can you imagine setting clocks through NNTP?

Start with posting a message to a newsgroup asking "what time is it?"

That will be followed by 5 contradictory answers and 3 entirely irrelevant responses. Each of those will each be responded to by 5 equally inaccurate flames.

Eventually, someone will mention in a posting that Hitler never wore a wristwatch, and the whole thing will be over.

No, you want an embedded module

[splorf]

Most GPS instruments (i.e. handheld gizmos with displays) have NMEA output, which gives the time and position through a serial port in ascii. This is very easy to interface to a computer (just plug it in and write some trivial decoding software) but woefully inaccurate (up to a substantial fraction of a second). That's partly because the serial port in a GPS is considered part of the user interface, whic h runs as a low priority task.

For accurate time you want a 1 PPS output. GPS chipsets and embedded modules (like that $24.95 thing that appears to be sold out) usually have that. The higher quality modules 1 PPS output is good to within a few microseconds, but the cheaper ones can be off by 25-50 microseconds.

Here is the famous W3IWI Totally Accurate Clock available in kit form from TAPR for the princely sum of $139.00 for non-TAPR-members. That's a pretty good deal--that $24.95 module was a stupendous deal and had to be some kind of surplus closeout.