Re: [ntp:questions] Jitter versus polling interval
Danny Mayer wrote: > David J Taylor wrote: >> David Woolley wrote: >>> Danny Mayer wrote: >>> Increased polling frequency increases jitter. >>> I think jitter was being used loosely here, not with the specific >>> sense of high frequency variations. >> >> Yes, my original question referred to looking at a graph of offset >> plotted over a day (sampling at 5-minute intervals) and trying to >> judge the amount of short-term oscillation as opposed to the >> longer-term drifts (e.g. those caused by temperature variations). > > Perform a FFT (Fast Fourier Transform) on the data and look for the > peaks. > > Danny A visual inspection is more than adequate in this case, although I agree that the FFT could be used for quantitative analysis. Cheers, David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] Increase maximum frequency offset to deal with really bad clock
Frank Wayne wrote: > While several people suggested that Jeff use guest/host synchronization, no > one ever answered the first question, "is it possible to override the maximum > offset PPM?" > > -Original Message- > From: questions-bounces+fwayne=frankwayne@lists.ntp.org > [mailto:questions-bounces+fwayne=frankwayne@lists.ntp.org] On Behalf Of > j...@sailorfej.net > Sent: Wednesday, 28 January 2009 14:46 > To: questions@lists.ntp.org > Subject: Increase maximum frequency offset to deal with really bad clock > > Hi folks, > > I have a some FreeBSD systems running as guests in Microsoft Virtual > Server. There is a known problem with these were the clocks run very > fast. I am trying to use ntpd to keep their clocks in sync, but the > frequency error offset is exceeding (I think) ntpd's maximum of 500, > my driftfile always contains a value of "-500.000". If understand the > way this works correctly, if I could get the frequency error offset to > represent the real error rate which I believe to be much higher that > 500 PPM, then ntpd would be able to keep the clocks in sync, as it is > now, it slowly falls behind until it fails to correct altogether. > > So is it possible to override the maximum offset PPM in the driftfile, > or is there a better way to fix this? You can re-centre the frequency error in 100ppm steps, using tickadj. However, for VMWare, at least, the effective frequency varies wildly over relatively short durations, as it attempts to maintain a correct long term tick rate. These are likely to confuse ntpd, whatever the frequency correction limits are set to. Before recompiling with a higher limit, one should also check the kernel limits; this limit may still apply if one doesn't use the kernel PLL. In ntpd V3, the 500ppm limit was actually set by the kernel, not by ntpd, and was less on some systems. It also used scaled integer arithmetic, which might have suffered overflows if some limits were changed - I'm not sure if ntpd V4 is completely immune from that. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] What is the "best" synchronization possible over the network?
eugene...@sbcglobal.net wrote: > On Feb 10, 5:30 pm, n...@tla.org (John Ioannidis) wrote: > > The problem setup: two locations, both within the United States, neither > > has roof access so no GPS reception is possible. How do you synchronize > > them with better than 50-microsecond accuracy? > > You might want to consider a CDMA receiver such as these units from > EndRun Technologies: > http://www.endruntechnologies.com/network-time-source.htm > http://www.endruntechnologies.com/network-time-server.htm I know of one installation using those and I noticed that they didn't support the recent leap second very well (they were off by a second until an admin did something days later). I don't know if it's caused by the device, the driver, ntpd, bad configuration or something else though. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] Reference implementation 4.2.4p6 receive() swapping bytes..?
regan_russ...@hotmail.com wrote: [] > My code (and project) works, I learnt a lot (a real lot believe me) > and I have to start on clean up, an installer and documentation, maybe > tomorrow. [] > Regan Regan, You might want to look at InnoSetup, in case you don't already know it. Cheers, David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] Very rapid polling
On Feb 9, 3:07 pm, jlevine wrote: > In the last few days I have seen an increasing number of systems that > are requesting the time in NTP format several times per second. This > poll interval is far in excess of the usual best practices. Since > there are a number of such systems, it is possible that this problem > is a result of a new version of NTP that has just been released. > Please let me know if you have any information about a new version of > NTP that can do this or if any of you are seeing the same problem. Over at the NTP Pool we see this quite a bit. It's typical that on any one server a couple of IPs (out of thousands) are responsible for several percent of the requests. Some clients will stop when the replies stop. Others are running at the crazy query rate because they don't get the replies (broken firewall configuration, for example). I don't think have ever gotten consensus on what strategy works the best. - ask ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] Windows ntpd highest priority doesn't require admin access
Danny Mayer wrote: [] > Why would the account even be in the adminstrators group? It shouldn't > even be in the users group since that also has too many privileges. It could be there because you have put it there during testing a new version, to allow change of priority. It's a lot quicker to do that than to use the policy editor, and it's also easy to change back. > There are only two privileges that should be required: Logon as > service and Change system time. If you want to add Increase > Scheduling priority you can if you think it's necessary. It's not > strictly necessary for ntp from what I have seen. > > Danny I have found that running at real-time makes little difference on a lightly loaded system running as a stratum 1 server, but it might help on a CPU-saturated system. Of course, I would not be expecting any great precision from such a loaded PC! David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] Reference implementation 4.2.4p6 receive() swapping bytes..?
Danny Mayer wrote: > Dave Hart wrote: [] > He didn't even mention Windows, why would you expect him to be using > Windows utilities? As far as I can tell you have made no patches > available against 4.2.4p6 and they don't belong there anyway. They > should be in the ntp-dev branch. > > Danny I, for once, certainly hope that Dave Harts improvements to NTP for Windows make it into a formal version as soon as possible. He has worked hard on the software, building on what has been done before, and it's now working much better as a result. I'm running it on all my Windows PCs. David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
David, David J Taylor wrote: > Richard B. Gilbert wrote: > [] >> I've been using Comcast for five or six years now without a problem! >> YMMV. > > There have been a few problems with my ISP, hence I moved to a 3rd party. We did't ever have any problems using the DNS servers of our ISPs. [...] > I used to have my own DNS server. You do /not/ have to get the rest of > the world from Comcast. DNS is /not/ like NTP, and you talk directly to > the root servers to start with, which then pass you down gradually to the > correct server for the domain (e.g. for microsoft.com) and in future you > query the relevant domain server directly. All seems to work very > smoothly and automatically. IMHO DNS is not like NTP in the sense that is just *starts* sending queries for domains which have not yet been resolved to the root DNS servers and then are redirected/go on down to the authoritative DNS for that domain, whereas NTP would stuck with the top level servers if they have been configured. However, if several local subnets needed to resolve "microsoft.com" then each one would have to ask the root servers the first time. If you send DNS queries to your ISP's servers then this would save bandwidth since they are normally closer to your network than the root DNS servers, and there is a chance that other customers of your ISP has already sent queries for "microsoft.com" so this has already been cached by the ISP's DNS servers and no request has to be made to the root servers. IMHO in this sense it's similar to NTP so the load is distributed and not concentrated on the root servers. Martin -- Martin Burnicki Meinberg Funkuhren Bad Pyrmont Germany ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Martin Burnicki wrote: > IMHO DNS is not like NTP in the sense that is just *starts* sending queries > for domains which have not yet been resolved to the root DNS servers and > then are redirected/go on down to the authoritative DNS for that domain, > whereas NTP would stuck with the top level servers if they have been > configured. > > However, if several local subnets needed to resolve "microsoft.com" then > each one would have to ask the root servers the first time. > > If you send DNS queries to your ISP's servers then this would save bandwidth > since they are normally closer to your network than the root DNS servers, Also don't forget that it will save time too, when you are on a connection with some roundtriptime. When you use the ISP forwarder, you make a single request to the forwarder, which will make the multiple requests required to do the lookup, presumably over a faster connection than you have yourself. It will then send a single reply with the results. When you go to the root server yourself, you have to do multiple queries over the slower connection, which takes more time. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Rob wrote: > Martin Burnicki wrote: >> IMHO DNS is not like NTP in the sense that is just *starts* sending >> queries for domains which have not yet been resolved to the root DNS >> servers and then are redirected/go on down to the authoritative DNS for >> that domain, whereas NTP would stuck with the top level servers if they >> have been configured. >> >> However, if several local subnets needed to resolve "microsoft.com" then >> each one would have to ask the root servers the first time. >> >> If you send DNS queries to your ISP's servers then this would save >> bandwidth since they are normally closer to your network than the root >> DNS servers, > > Also don't forget that it will save time too, when you are on a connection > with some roundtriptime. > > When you use the ISP forwarder, you make a single request to the > forwarder, which will make the multiple requests required to do the > lookup, presumably > over a faster connection than you have yourself. It will then send a > single reply with the results. > > When you go to the root server yourself, you have to do multiple queries > over the slower connection, which takes more time. Agreed, this is also a good point. Martin -- Martin Burnicki Meinberg Funkuhren Bad Pyrmont Germany ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
Jack, jack wrote: > Hi all, > > I am trying to sync my Windows box to an external GPS source. I > currently have BU353, whose output is not very periodic. I read up on > ntpd implementation that uses PPS signal but I don't even have an RS > 232 port on my computer. > > My questions: > 1) what's the best GPS antenna (and protocol) in terms of > consistencies in its output? > 2) what kind of accuracy can i expect? This is somewhat limited under Windows due to the granularity of the Windows system clock. Alternatively you could install a GPS PCI card which your applications can also access directly, so you can benefit from the full accuracy of the card, e.g.: http://www.meinberg.de/english/products/gps170pex.htm Take care, though, I'm biased! ;-)) Martin -- Martin Burnicki Meinberg Funkuhren Bad Pyrmont Germany ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
David Woolley wrote: > Unruh wrote: > >> >> No. That is what GPS does for you. It determines your postion, then uses >> that to determine the time delay from the sattelite to your receiver. > > Actually, it is the other way round. It uses the differences in the > time delays to solve for the position (and the actual time delays). Right. However, it computes the position of the *antenna*. So if you receiver is connected via a long antenna cable you have to compensate the signal delay on the cable. This can not be done automatically by the GPS receiver. For examples, for our GPS receivers the antenna can be installed up to 300 meters away from the receiver, and you can configure the antenna cable length in the receiver to compensate that delay. Martin -- Martin Burnicki Meinberg Funkuhren Bad Pyrmont Germany ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Martin Burnicki wrote: > David, > > David J Taylor wrote: >> Richard B. Gilbert wrote: >> [] >>> I've been using Comcast for five or six years now without a problem! >>> YMMV. >> >> There have been a few problems with my ISP, hence I moved to a 3rd >> party. > > We did't ever have any problems using the DNS servers of our ISPs. At the time, mine was using servers in the USA (from the UK) and via non-reciprocal paths. Even now, it seems to be using servers from abroad, and has no local reference clock I don't think that anyone "cared" for it. >> I used to have my own DNS server. You do /not/ have to get the rest >> of the world from Comcast. DNS is /not/ like NTP, and you talk >> directly to the root servers to start with, which then pass you down >> gradually to the correct server for the domain (e.g. for >> microsoft.com) and in future you query the relevant domain server >> directly. All seems to work very smoothly and automatically. > > IMHO DNS is not like NTP in the sense that is just *starts* sending > queries for domains which have not yet been resolved to the root DNS > servers and then are redirected/go on down to the authoritative DNS > for that domain, whereas NTP would stuck with the top level servers > if they have been configured. Agreed. When I first answered I had missed that Richard probably knows more about DNS than me! > However, if several local subnets needed to resolve "microsoft.com" > then each one would have to ask the root servers the first time. Wouldn't you have one or two central DNS servers for both subnets? > If you send DNS queries to your ISP's servers then this would save > bandwidth since they are normally closer to your network than the > root DNS servers, and there is a chance that other customers of your > ISP has already sent queries for "microsoft.com" so this has already > been cached by the ISP's DNS servers and no request has to be made to > the root servers. > > IMHO in this sense it's similar to NTP so the load is distributed and > not concentrated on the root servers. > > Martin Agreed, it's a distributed system. I monitor the faults reported to my ISP, and DNS problems do come up, hence my preference for a 3rd party solution. It's also one less thing to change were I to change ISP. Cheers, David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
David J Taylor wrote: > Martin Burnicki wrote: >> However, if several local subnets needed to resolve "microsoft.com" >> then each one would have to ask the root servers the first time. > > Wouldn't you have one or two central DNS servers for both subnets? Sorry, I've been not specific enough here. I meant e.g. several companies who have their own local subnets and local DNS servers. If they don't use their ISP's servers then the local DNS server in every company would contact the root DNS servers directly. If they used their ISP's DNS servers then all customers of the same ISP simply get the cached results for popular domains without having to contact the root servers. Of cause all of this discussion is based on assumptions, how much traffic and load on the root servers would differ if people used them rather then their ISP's servers. Martin -- Martin Burnicki Meinberg Funkuhren Bad Pyrmont Germany ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
[ntp:questions] Jupiter refclock on Windows
In theory, my recent test releases include support for Jupiter reference clocks connected via serial with PPS. I don't have a Jupiter to test against, but if you have a Windows machine with a serial port and a Jupiter receiver, you could give it a whirl. The Jupiter driver requires PPSAPI, so you must install serialpps.sys to use it on Windows: http://davehart.net/ntp/refclock/serialpps-20090301.zip http://davehart.net/ntp/refclock/ntp-4.2.4p6-DLH-QPC-20090306-bin.zip The easiest way to get going if you haven't been running NTP on that machine is to use Meinberg's installer, then replace the ntpd.exe or all the binaries from the -bin.zip. Contact me with any questions/ problems. Cheers, Dave Hart ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Martin Burnicki wrote: > David J Taylor wrote: >> Martin Burnicki wrote: >>> However, if several local subnets needed to resolve "microsoft.com" >>> then each one would have to ask the root servers the first time. >> >> Wouldn't you have one or two central DNS servers for both subnets? > > Sorry, I've been not specific enough here. > > I meant e.g. several companies who have their own local subnets and > local DNS servers. > > If they don't use their ISP's servers then the local DNS server in > every company would contact the root DNS servers directly. > > If they used their ISP's DNS servers then all customers of the same > ISP simply get the cached results for popular domains without having > to contact the root servers. > > Of cause all of this discussion is based on assumptions, how much > traffic and load on the root servers would differ if people used them > rather then their ISP's servers. > > > Martin Martin, When I was involved in this, some years ago now, each site had its own DNS servers, and those spoke to a central company-wide DNS server. The connection to the ISP was central, and not per-site. I suppose that, like NTP, DNS should be configured to best suit local needs, and try to place minimum load on central servers, although the requirements and resulting configuration may be different. Cheers, David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Martin Burnicki wrote: > David, > > David J Taylor wrote: >> Richard B. Gilbert wrote: >> [] >>> I've been using Comcast for five or six years now without a problem! >>> YMMV. >> There have been a few problems with my ISP, hence I moved to a 3rd party. > > We did't ever have any problems using the DNS servers of our ISPs. > > [...] >> I used to have my own DNS server. You do /not/ have to get the rest of >> the world from Comcast. DNS is /not/ like NTP, and you talk directly to >> the root servers to start with, which then pass you down gradually to the >> correct server for the domain (e.g. for microsoft.com) and in future you >> query the relevant domain server directly. All seems to work very >> smoothly and automatically. > > IMHO DNS is not like NTP in the sense that is just *starts* sending queries > for domains which have not yet been resolved to the root DNS servers and > then are redirected/go on down to the authoritative DNS for that domain, > whereas NTP would stuck with the top level servers if they have been > configured. > > However, if several local subnets needed to resolve "microsoft.com" then > each one would have to ask the root servers the first time. > > If you send DNS queries to your ISP's servers then this would save bandwidth > since they are normally closer to your network than the root DNS servers, > and there is a chance that other customers of your ISP has already sent > queries for "microsoft.com" so this has already been cached by the ISP's > DNS servers and no request has to be made to the root servers. > > IMHO in this sense it's similar to NTP so the load is distributed and not > concentrated on the root servers. > > Martin It is nothing like NTP. You cannot make that comparison. You are not really saving anything in the way of bandwidth or speed and almost no query from different groups of people is going to have an answer already cached at the ISP's DNS. It's a lot like using an HTTP Caching Proxy: people thought it would be beneficial but it was quickly shown that less than 10% of the pages were already cached and that number was decreasing rapidly. Jeff Mogul wrote a paper on that a number of years ago. Danny -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Rob wrote: > Martin Burnicki wrote: >> IMHO DNS is not like NTP in the sense that is just *starts* sending queries >> for domains which have not yet been resolved to the root DNS servers and >> then are redirected/go on down to the authoritative DNS for that domain, >> whereas NTP would stuck with the top level servers if they have been >> configured. >> >> However, if several local subnets needed to resolve "microsoft.com" then >> each one would have to ask the root servers the first time. >> >> If you send DNS queries to your ISP's servers then this would save bandwidth >> since they are normally closer to your network than the root DNS servers, > > Also don't forget that it will save time too, when you are on a connection > with some roundtriptime. > > When you use the ISP forwarder, you make a single request to the forwarder, > which will make the multiple requests required to do the lookup, presumably > over a faster connection than you have yourself. It will then send a single > reply with the results. > > When you go to the root server yourself, you have to do multiple queries > over the slower connection, which takes more time. Have you ever tested this? Tests show that this is almost never the case and the time saved is so small you will never notice the difference. Danny -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
David J Taylor wrote: > Martin Burnicki wrote: >> David J Taylor wrote: >>> Martin Burnicki wrote: However, if several local subnets needed to resolve "microsoft.com" then each one would have to ask the root servers the first time. >>> Wouldn't you have one or two central DNS servers for both subnets? >> Sorry, I've been not specific enough here. >> >> I meant e.g. several companies who have their own local subnets and >> local DNS servers. >> >> If they don't use their ISP's servers then the local DNS server in >> every company would contact the root DNS servers directly. >> >> If they used their ISP's DNS servers then all customers of the same >> ISP simply get the cached results for popular domains without having >> to contact the root servers. >> >> Of cause all of this discussion is based on assumptions, how much >> traffic and load on the root servers would differ if people used them >> rather then their ISP's servers. >> >> >> Martin > > Martin, > > When I was involved in this, some years ago now, each site had its own DNS > servers, and those spoke to a central company-wide DNS server. The > connection to the ISP was central, and not per-site. > > I suppose that, like NTP, DNS should be configured to best suit local > needs, and try to place minimum load on central servers, although the > requirements and resulting configuration may be different. > You can't actually accomplish that, you just increase the load on the ISP's servers. Danny > Cheers, > David -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
"David J Taylor" wrote in message news:jzqtl.5203$lc7.2...@text.news.virginmedia.com... > Martin Burnicki wrote: [...] >> We did't ever have any problems using the DNS servers of our ISPs. > > At the time, mine was using servers in the USA (from the UK) and via > non-reciprocal paths. Even now, it seems to be using servers from > abroad, and has no local reference clock I don't think that anyone > "cared" for it. I think you're talking about your ISP's _NTP_ server(s) here. Groetjes, Maarten Wiltink ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Danny Mayer wrote: > Have you ever tested this? Tests show that this is almost never the case > and the time saved is so small you will never notice the difference. I certainly won't take anything you post as a fact, be it only because you write in such negative tone and pretend to know everything better. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] Making a clock tell the wrong time?
> > For testing purposes I want to configure an NTP server to run with a > > small but known offset. I would like to test at the following offset > > values 25, 90, 180, 350 and 2500ms. > > What is this test intended to prove? I want to understand how long it takes in practice for an offset to be propagated from stratum 1 servers through a set of stratum 2 servers to a stratum 3 client. I want to see how the poll intervals vary at stratum 2 and 3 in response to changing time. I want to verify that a monitoring system which is meant to raise an alarm when a clock shows an offset greeater than a particular threshold, does in fact raise that alarm. This is on an isolated lab network. There are real stratum 1 servers available, synced to GPS as well as my pseudo-stratum 1 test server. By using multiple virtual IP addresses I can set the test server to emulate a majority of clocks so the stratum 2 servers will prefer the time from my test clock. > Setting the Undiscplined Local Clock to stratum 0 instead of the default > stratum 5 does not make it a better "clock". What are you trying to > accomplish with this? When my test clock is sitting alongside one or more genuine stratum 1 clocks I want to be sure that the time it presents is treated equally by the clock selection and filtering algorithms. It would be treated differently if it ran at stratum 5 alongside a set of stratum 1s. > > with a fudge line to set the time1 parameter to 90ms. > > > The server showed a 90ms offset after the first poll of the local > > clock, but within a few poll cycles it had discarded the offset. > > That's because ntpd disciplined the clock to amortize that offset. > > Once this has happened, the clock is now operating at the 90ms offset. > > Isn't this what you wanted? What you describe is what I expected. What I wanted was for that offset to be reflected in the time output by the ntpd. What I saw was a very slow slew, which generated about 3ms offset in an hour - this is just under 1ppm which I would expect from a free- running local clock which has a fairly current drift correction.. When you talk about the offset being amortized, do you mean that it has passed to the kernel and the kernel then applies a slew to make the change? I repeated the test with an offset of 180ms in the hope that instead of seeing a slew I would see a step in the output time from ntpd (after the stepout threshold had elapsed), but again the ntpd just showed a very small offset. It may well be that this is normal behaviour for the local clock driver. I am now working to configure a serial input to this PC which I think will use the fudge value in the way that I want. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Rob wrote: > Danny Mayer wrote: >> Have you ever tested this? Tests show that this is almost never the case >> and the time saved is so small you will never notice the difference. > > I certainly won't take anything you post as a fact, be it only because you > write in such negative tone and pretend to know everything better. After 10 years doing DNS development, I should know something about the subject. It would take time do dig up the pointers to the work people have done on this. Danny -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
David J Taylor wrote: > Martin Burnicki wrote: >> David, >> >> David J Taylor wrote: >>> Richard B. Gilbert wrote: >>> [] I've been using Comcast for five or six years now without a problem! YMMV. >>> >>> There have been a few problems with my ISP, hence I moved to a 3rd >>> party. >> >> We did't ever have any problems using the DNS servers of our ISPs. > > At the time, mine was using servers in the USA (from the UK) and via > non-reciprocal paths. Even now, it seems to be using servers from > abroad, and has no local reference clock I don't think that anyone > "cared" for it. > >>> I used to have my own DNS server. You do /not/ have to get the rest >>> of the world from Comcast. DNS is /not/ like NTP, and you talk >>> directly to the root servers to start with, which then pass you down >>> gradually to the correct server for the domain (e.g. for >>> microsoft.com) and in future you query the relevant domain server >>> directly. All seems to work very smoothly and automatically. >> >> IMHO DNS is not like NTP in the sense that is just *starts* sending >> queries for domains which have not yet been resolved to the root DNS >> servers and then are redirected/go on down to the authoritative DNS >> for that domain, whereas NTP would stuck with the top level servers >> if they have been configured. > > Agreed. When I first answered I had missed that Richard probably knows > more about DNS than me! > I used to be an "expert" but time has taken its toll; I have forgotten a lot and things have changed in the last ten years. Eleven years ago, my then employer's "upstream provider" asked us to stop using their DNS server. Fair enough; we had something like 2,000 student users! They sent and received a lot of mail; it was a heavy load! I set up a local DNS server. At first it was simply a "caching" server; ask once and keep the answer on hand. Then I built the database to make the server "authoritative" for the site, got our upstream provider to host a copy. . . . It was highly educational! ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Danny Mayer wrote: > David J Taylor wrote: [] >> Martin, >> >> When I was involved in this, some years ago now, each site had its >> own DNS servers, and those spoke to a central company-wide DNS >> server. The connection to the ISP was central, and not per-site. >> >> I suppose that, like NTP, DNS should be configured to best suit local >> needs, and try to place minimum load on central servers, although the >> requirements and resulting configuration may be different. >> > > You can't actually accomplish that, you just increase the load on the > ISP's servers. > > Danny ISP servers were not used. David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Maarten Wiltink wrote: > "David J Taylor" > wrote in > message news:jzqtl.5203$lc7.2...@text.news.virginmedia.com... >> Martin Burnicki wrote: > [...] >>> We did't ever have any problems using the DNS servers of our ISPs. >> >> At the time, mine was using servers in the USA (from the UK) and via >> non-reciprocal paths. Even now, it seems to be using servers from >> abroad, and has no local reference clock I don't think that >> anyone "cared" for it. > > I think you're talking about your ISP's _NTP_ server(s) here. > > Groetjes, > Maarten Wiltink Sorry for the confusion, but yes, I was. Having had problems with the NTP servers, I had no hesitation in moving to separate DNS servers when my ISP started having trouble there as well. Now I have my own simple stratum 1 NTP server. Cheers, David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] Making a clock tell the wrong time?
On 2009-03-10, Mike K Smith wrote: >Steve Kostecke wrote: > >>Mike K Smith wrote: >> >>> For testing purposes I want to configure an NTP server to run with >>> a small but known offset. I would like to test at the following >>> offset values 25, 90, 180, 350 and 2500ms. >> >> What is this test intended to prove? > > I want to understand how long it takes in practice for an offset to be > propagated from stratum 1 servers through a set of stratum 2 servers > to a stratum 3 client. [snip] >>> with a fudge line to set the time1 parameter to 90ms. >> >>> The server showed a 90ms offset after the first poll of the local >>> clock, but within a few poll cycles it had discarded the offset. >> >> That's because ntpd disciplined the clock to amortize that offset. >> >> Once this has happened, the clock is now operating at the 90ms offset. >> >> Isn't this what you wanted? > > What you describe is what I expected. What I wanted was for that > offset to be reflected in the time output by the ntpd. Now that I thing about it, what happened when you applied that offset is that you "moved" the clock 90ms. > What I saw was a very slow slew, which generated about 3ms offset in > an hour - this is just under 1ppm which I would expect from a free- > running local clock which has a fairly current drift correction.. The place to see view offset is from the client. Is that where you were looking from? -- Steve Kostecke NTP Public Services Project - http://support.ntp.org/ ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
Hi everyone, First of all, I thank you for all your response. I understand the best solution would be using an IRIG board and that's what we had been using. We are now trying to make our product more compact by using a small single board PC with no RS 232 or PCI slot (no slot at all). Currently I am using GPS antenna with a USB connection. I also understand that if I can get a small PC with one genuine RS232, I can use Dave Hart's solution as tested by David. I will try to my best to steer the project towards this direction. Right now, I am trying to find the best GPS antenna that outputs at a consistent pace (better if the output can be configured so I only receive time stamps). Any recommendations wrt that? Thanks again. Jack On Mar 10, 4:59 am, Martin Burnicki wrote: > Jack, > > jack wrote: > > Hi all, > > > I am trying to sync my Windows box to an external GPS source. I > > currently have BU353, whose output is not very periodic. I read up on > > ntpd implementation that uses PPS signal but I don't even have an RS > > 232 port on my computer. > > > My questions: > > 1) what's the best GPS antenna (and protocol) in terms of > > consistencies in its output? > > 2) what kind of accuracy can i expect? > > This is somewhat limited under Windows due to the granularity of the Windows > system clock. > > Alternatively you could install a GPS PCI card which your applications can > also access directly, so you can benefit from the full accuracy of the > card, e.g.:http://www.meinberg.de/english/products/gps170pex.htm > > Take care, though, I'm biased! ;-)) > > Martin > -- > Martin Burnicki > > Meinberg Funkuhren > Bad Pyrmont > Germany ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
jack wrote: > Hi everyone, > > First of all, I thank you for all your response. > > I understand the best solution would be using an IRIG board and that's > what we had been using. We are now trying to make our product more > compact by using a small single board PC with no RS 232 or PCI slot > (no slot at all). Currently I am using GPS antenna with a USB > connection. > > I also understand that if I can get a small PC with one genuine RS232, > I can use Dave Hart's solution as tested by David. I will try to my > best to steer the project towards this direction. > > Right now, I am trying to find the best GPS antenna that outputs at a > consistent pace (better if the output can be configured so I only > receive time stamps). Any recommendations wrt that? > > Thanks again. > > Jack > You might want to take a look at the work of Poul-Henning Kamp (PHK). He used Soekris 4501 single board computers together with a GPS receiver to make a highly accurate clock. This was long enough ago that most of the hardware he used is no longer available new but it might give you some ideas. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
jack wrote: > Hi everyone, > > First of all, I thank you for all your response. > > I understand the best solution would be using an IRIG board and that's > what we had been using. We are now trying to make our product more > compact by using a small single board PC with no RS 232 or PCI slot > (no slot at all). Currently I am using GPS antenna with a USB > connection. > > I also understand that if I can get a small PC with one genuine RS232, > I can use Dave Hart's solution as tested by David. I will try to my > best to steer the project towards this direction. > > Right now, I am trying to find the best GPS antenna that outputs at a > consistent pace (better if the output can be configured so I only > receive time stamps). Any recommendations wrt that? > > Thanks again. > > Jack Jack, GPS18 LVC https://buy.garmin.com/shop/shop.do?pID=223 GPS18x LVC https://buy.garmin.com/shop/shop.do?cID=158&pID=27594 The newer "x" version appears to be a lot more sensitive, at the expense of somewhat higher current consumption. You can tell the LVC version what sentences to send. Better check the technical spec to see what the USB one can be told to do: http://www.garmin.com/manuals/GPS18x_TechnicalSpecifications.pdf As others have said, the single-board PC with FreeBSD would probably be the best solution. I would go for that, except I happen to have a Windows PC which will do well enough for my purposes (say: within a millisecond). I've even heard that you could perhaps use a router as a compact Linux PC, and use the serial port which is either available on the box, or as a header on the PCB. Cheers, David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
On Mar 10, 3:44 pm, "Richard B. Gilbert" wrote: > You might want to take a look at the work of Poul-Henning Kamp (PHK). > He used Soekris 4501 single board computers together with a GPS receiver > to make a highly accurate clock. This was long enough ago that most of > the hardware he used is no longer available new but it might give you > some ideas. Seconded. FreeBSD will give you better timekeeping than Windows. If the difference between a few microseconds and a few hundred microseconds is no problem, my recent Windows releases are fine. But if you're dedicating a small PC to the task, care about those fractions of a millisecond, and can manage FreeBSD, it's the gold standard for NTP. Cheers, Dave Hart ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] Making a clock tell the wrong time?
On 10 Mar, 14:29, Steve Kostecke wrote: > On 2009-03-10, Mike K Smith wrote: > > >Steve Kostecke wrote: > > >>Mike K Smith wrote: > Now that I thing about it, what happened when you applied that offset is > that you "moved" the clock 90ms. > > > What I saw was a very slow slew, which generated about 3ms offset in > > an hour - this is just under 1ppm which I would expect from a free- > > running local clock which has a fairly current drift correction.. > > The place to see view offset is from the client. Is that where you were > looking from? Yes. The client is referenced to the genuine stratum 1 servers. I have now reconfigured the PC to take a serial feed (NMEA) and the use of the fudged time1 value on the NMEA driver is now working exactly as I expected. I have removed the local refclock from the config as it is of no value. Thanks for following up on this. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
[ntp:questions] micro-optimization
I have released a new test version of 4.2.4p6 with numerous Windows- specific improvements compared to the baseline 4.2.4p6. Since my last release, the most significant change is to read the processor cycle counter using the RDTSC instruction directly when it is equivalent to QueryPerformanceCounter. When it is not equivalent, ntpd is allowed to roam freely across all logical processors once again. I've also made small changes to the counter frequency tuning code to balance observation time (longer is better quality, to a point, but also means it takes longer to effect the counter frequency compensation after startup). The counter frequency tuning messages have been further quieted to once every 8 hours after the initial compensation is done. http://davehart.net/refclock/ has binaries, debug binaries, symbols, and source, most people will want: http://davehart.net/ntp/refclock/ntp-4.2.4p6-DLH-QPC-20090310-bin.zip http://tinyurl.com/cz8kk4 Cheers, Dave Hart ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
I cannot switch to Linux because my main application runs only on Windows. It would take us a long time to port to Linux. If I can get to millisecond accuracy, I'd be happy. Jack On Mar 10, 11:56 am, Dave Hart wrote: > On Mar 10, 3:44 pm, "Richard B. Gilbert" > wrote: > > > You might want to take a look at the work of Poul-Henning Kamp (PHK). > > He used Soekris 4501 single board computers together with a GPS receiver > > to make a highly accurate clock. This was long enough ago that most of > > the hardware he used is no longer available new but it might give you > > some ideas. > > Seconded. FreeBSD will give you better timekeeping than Windows. If > the difference between a few microseconds and a few hundred > microseconds is no problem, my recent Windows releases are fine. But > if you're dedicating a small PC to the task, care about those > fractions of a millisecond, and can manage FreeBSD, it's the gold > standard for NTP. > > Cheers, > Dave Hart ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
jack wrote: > I cannot switch to Linux because my main application runs only on > Windows. It would take us a long time to port to Linux. have you looked at running under wine or compiling a native linux app against winelib? http://www.winehq.org/about/ uwe ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
Martin Burnicki writes: >David Woolley wrote: >> Unruh wrote: >> >>> >>> No. That is what GPS does for you. It determines your postion, then uses >>> that to determine the time delay from the sattelite to your receiver. >> >> Actually, it is the other way round. It uses the differences in the >> time delays to solve for the position (and the actual time delays). Actaully they solve for the position and the time delay all at once. But if the receiver knows its position, it only needs one sattelite to determine the time, and it uses the position to determine the delay. >Right. However, it computes the position of the *antenna*. So if you >receiver is connected via a long antenna cable you have to compensate the >signal delay on the cable. This can not be done automatically by the GPS >receiver. Agreed. On the other hand, 300m is 1usec and most computers are completely incapable of discipling the clock to that accuracy ( interrupt latencies, etc) If you actually have a system which could do better than that, then antenna delay is important. As may proper cable termination so the signal bouncing back and forth along the cable does not mess up the timing. >For examples, for our GPS receivers the antenna can be installed up to 300 >meters away from the receiver, and you can configure the antenna cable >length in the receiver to compensate that delay. >Martin >-- >Martin Burnicki >Meinberg Funkuhren >Bad Pyrmont >Germany ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
"David J Taylor" writes: >jack wrote: >> Hi everyone, >> >> First of all, I thank you for all your response. >> >> I understand the best solution would be using an IRIG board and that's >> what we had been using. We are now trying to make our product more >> compact by using a small single board PC with no RS 232 or PCI slot >> (no slot at all). Currently I am using GPS antenna with a USB >> connection. >> >> I also understand that if I can get a small PC with one genuine RS232, >> I can use Dave Hart's solution as tested by David. I will try to my >> best to steer the project towards this direction. >> >> Right now, I am trying to find the best GPS antenna that outputs at a >> consistent pace (better if the output can be configured so I only >> receive time stamps). Any recommendations wrt that? >> >> Thanks again. >> >> Jack >Jack, >GPS18 LVC > https://buy.garmin.com/shop/shop.do?pID=223 >GPS18x LVC > https://buy.garmin.com/shop/shop.do?cID=158&pID=27594 >The newer "x" version appears to be a lot more sensitive, at the expense >of somewhat higher current consumption. You can tell the LVC version what >sentences to send. Better check the technical spec to see what the USB >one can be told to do: > http://www.garmin.com/manuals/GPS18x_TechnicalSpecifications.pdf Unfortnately, the LVC is a serial port output, plus a PPS output. The USB version has no PPS output and is liable to have pretty bad timing (where good=usec). >As others have said, the single-board PC with FreeBSD would probably be >the best solution. I would go for that, except I happen to have a Windows >PC which will do well enough for my purposes (say: within a millisecond). >I've even heard that you could perhaps use a router as a compact Linux PC, >and use the serial port which is either available on the box, or as a >header on the PCB. >Cheers, >David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
On Tue, 10 Mar 2009 09:17:58 -0700 (PDT), jack wrote: > I cannot switch to Linux because my main application runs only on > Windows. It would take us a long time to port to Linux. > If I can get to millisecond accuracy, I'd be happy. > > Jack > > On Mar 10, 11:56 am, Dave Hart wrote: >> On Mar 10, 3:44 pm, "Richard B. Gilbert" >> wrote: >> >> > You might want to take a look at the work of Poul-Henning Kamp (PHK). >> > He used Soekris 4501 single board computers together with a GPS receiver >> > to make a highly accurate clock. This was long enough ago that most of >> > the hardware he used is no longer available new but it might give you >> > some ideas. >> >> Seconded. FreeBSD will give you better timekeeping than Windows. If >> the difference between a few microseconds and a few hundred >> microseconds is no problem, my recent Windows releases are fine. But >> if you're dedicating a small PC to the task, care about those >> fractions of a millisecond, and can manage FreeBSD, it's the gold >> standard for NTP. >> >> Cheers, >> Dave Hart FreeBSD is not Linux. Certainly not when talking timing. /hjj > ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
jack writes: >Hi everyone, >First of all, I thank you for all your response. >I understand the best solution would be using an IRIG board and that's >what we had been using. We are now trying to make our product more >compact by using a small single board PC with no RS 232 or PCI slot >(no slot at all). Currently I am using GPS antenna with a USB >connection. >I also understand that if I can get a small PC with one genuine RS232, >I can use Dave Hart's solution as tested by David. I will try to my >best to steer the project towards this direction. >Right now, I am trying to find the best GPS antenna that outputs at a >consistent pace (better if the output can be configured so I only >receive time stamps). Any recommendations wrt that? Not at all sure what you mean. IF you do not use the PPS output, you will never get timing better than a few ms. Not at all sure what :consistant pace" means. And from what I have read, USB itself has its own variability. Ie, you cannot get consistant output from a USB port, no matter what receiver you use. It is like giving your secretary a highly accurate deskclock, and then getting the time by telling her to read it and walk over to your office on another floor to tell you the time. >Thanks again. >Jack >On Mar 10, 4:59=A0am, Martin Burnicki >wrote: >> Jack, >> >> jack wrote: >> > Hi all, >> >> > I am trying to sync my Windows box to an external GPS source. I >> > currently have BU353, whose output is not very periodic. I read up on >> > ntpd implementation that uses PPS signal but I don't even have an RS >> > 232 port on my computer. >> >> > My questions: >> > 1) what's the best GPS antenna (and protocol) in terms of >> > consistencies in its output? >> > 2) what kind of accuracy can i expect? >> >> This is somewhat limited under Windows due to the granularity of the Wind= >ows >> system clock. >> >> Alternatively you could install a GPS PCI card which your applications ca= >n >> also access directly, so you can benefit from the full accuracy of the >> card, e.g.:http://www.meinberg.de/english/products/gps170pex.htm >> >> Take care, though, I'm biased! =A0;-)) >> >> Martin >> -- >> Martin Burnicki >> >> Meinberg Funkuhren >> Bad Pyrmont >> Germany ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
Unruh wrote: [] >> The newer "x" version appears to be a lot more sensitive, at the >> expense of somewhat higher current consumption. You can tell the >> LVC version what sentences to send. Better check the technical spec >> to see what the USB one can be told to do: >> http://www.garmin.com/manuals/GPS18x_TechnicalSpecifications.pdf > > Unfortnately, the LVC is a serial port output, plus a PPS output. The > USB > version has no PPS output and is liable to have pretty bad timing > (where good=usec). .. which is why I suggested checking in the manual as to whether the output could be set to be a better determined length, which is what the OP requested. It may be he /has/ to use USB. By the way, while timing on USB may not be wonderful, I have an application receiving some 40GB per day over USB, in fairly small packets from a DVB satellite feed, and the buffering isn't that great (a few kilobytes, I believe). Most often it's lossless, but it may be that the error-correction in the stream-to-file software does a good job! Cheers, David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
Dave Hart wrote: > > You might want to spend a little time curling up with RFC 3484, > "Default Address Selection for IPv6" > > http://www.rfc-editor.org/rfc/rfc3484.txt > > With RFC 3484 support, getaddrinfo sorts its results so that > applications processinig the results in order follow the selected > policy. Given that the RFC came out of Microsoft Research, it should > be no surprise that a certain widely-used platform respects RFC 3484. > Take particular note of the policy tables described in RFC 3484 and > how they allow site policies to come into play. > >From what I have experienced, this RFC has been an utter disaster -- it has destroyed the commonly used practice of load balancing by DNS round-robin. -hpa ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
David, Thanks again for your suggestion. It seems Garmin 18X USB can be tailored to only output certain messages as you said. Hopefully this will improve the performance. I am going to give it a try. To all others: I am only trying to find the best solution given my other constraints. Will try in the future to go with a system that can take advantage of the 1PPS signal. jack On Mar 10, 2:18 pm, "David J Taylor" wrote: > Unruh wrote: > > [] > > >> The newer "x" version appears to be a lot more sensitive, at the > >> expense of somewhat higher current consumption. You can tell the > >> LVC version what sentences to send. Better check the technical spec > >> to see what the USB one can be told to do: > >> http://www.garmin.com/manuals/GPS18x_TechnicalSpecifications.pdf > > > Unfortnately, the LVC is a serial port output, plus a PPS output. The > > USB > > version has no PPS output and is liable to have pretty bad timing > > (where good=usec). > > .. which is why I suggested checking in the manual as to whether the > output could be set to be a better determined length, which is what the OP > requested. It may be he /has/ to use USB. > > By the way, while timing on USB may not be wonderful, I have an > application receiving some 40GB per day over USB, in fairly small packets > from a DVB satellite feed, and the buffering isn't that great (a few > kilobytes, I believe). Most often it's lossless, but it may be that the > error-correction in the stream-to-file software does a good job! > > Cheers, > David ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] http://www.ntp.org/ => a blank page?
H. Peter Anvin wrote: > Dave Hart wrote: >> You might want to spend a little time curling up with RFC 3484, >> "Default Address Selection for IPv6" >> >> http://www.rfc-editor.org/rfc/rfc3484.txt >> >> With RFC 3484 support, getaddrinfo sorts its results so that >> applications processinig the results in order follow the selected >> policy. Given that the RFC came out of Microsoft Research, it should >> be no surprise that a certain widely-used platform respects RFC 3484. >> Take particular note of the policy tables described in RFC 3484 and >> how they allow site policies to come into play. >> > >>From what I have experienced, this RFC has been an utter disaster -- it > has destroyed the commonly used practice of load balancing by DNS > round-robin. > > -hpa Agreed. See Ask's and my comments in namedroppers. Danny -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
On 2009-03-10, jack wrote: > It seems Garmin 18X USB can be tailored to only output certain > messages as you said. Hopefully this will improve the performance. I > am going to give it a try. Just keep in mind that the GPS sends the NMEA sentence(s) when it is not busy doing other things. -- Steve Kostecke NTP Public Services Project - http://support.ntp.org/ ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] best gps receiver for time synchronization
jack wrote: > I cannot switch to Linux because my main application runs only on > Windows. It would take us a long time to port to Linux. > If I can get to millisecond accuracy, I'd be happy. You won't get millisecond accuracy on Windows. Although the software clock can be disciplined to better than a millisecond, applications can only read to one tick, which is 10ms by default and 1ms with the fastest multi-media timers (which risks lost ticks). ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] Making a clock tell the wrong time?
Mike K Smith wrote: > I want to understand how long it takes in practice for an offset to be > propagated from stratum 1 servers through a set of stratum 2 servers > to a stratum 3 client. > That's simple: 1) it should never happen - if it does the stratum one servers is badly broken; 2) if it does happen, the error shouldn't propagate, because the configuration should have enough good servers to outvote it. It's a garbage in case. A more reaslistic situation is that a client loses 10ms at frequent intervals during high disk activity, but only in 10ms steps. ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] Proposed NTP solution for a network
On Mar 4, 5:49 am, Terje Mathisen <"terje.mathisen at tmsw.no"> wrote:
> It is probably better than the Garmin, but almost certainly much less
> good than a dedicated FreeBSD box with an Oncore timing receiver.
>
> The most important point though is the fact that no such appliance box,
> (afaik, and I own/operate 3-4 different brands), supports proper logging
> of its own performance, so you have no way to check/verify after the
> fact if a given timestamp actually was good or not.
The Meinberg servers ("appliances") make available the loopstats
file. (I've no idea if that's "proper logging").
It generates graphs from them, too. An example here:
http://tmp.askask.com/2009/03/loopstat9669.gif
In my limited experience their support is really very excellent; I'd
recommend talking to them for advice. http://www.meinberg.de/english/
- ask
___
questions mailing list
questions@lists.ntp.org
https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] micro-optimization
On Mar 10, 11:13 am, Dave Hart wrote: > > Since my last > release, the most significant change is to read the processor cycle > counter using the RDTSC instruction directly when it is equivalent to > QueryPerformanceCounter. Just keep in mind that Windows' power management will change the frequency of a processor, thereby making the count returned by RDTSC have a variable rate on some processors (namely Intel's Pentium4 and AMD's Athlon and Opteron). HTH ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
Re: [ntp:questions] micro-optimization
On Mar 11, 2:01 am, Augustine wrote: > On Mar 10, 11:13 am, Dave Hart wrote: > > > > > Since my last > > release, the most significant change is to read the processor cycle > > counter using the RDTSC instruction directly when it is equivalent to > > QueryPerformanceCounter. > > Just keep in mind that Windows' power management will change the > frequency of a processor, thereby making the count returned by RDTSC > have a variable rate on some processors (namely Intel's Pentium4 and > AMD's Athlon and Opteron). I suspect those systems' HALs don't use RDTSC for QueryPerformanceCounter, so RDTSC won't be used. Only if they're found to be beating at the same rate (possibly with an offset) does RDTSC get substituted. Cheers, Dave Hart ___ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
