Re: [ntp:questions] How should an NTP server fail?
On Fri, Jun 11, 2010 at 03:08:53AM +, David L. Mills wrote: An hour later I disconnected the server. By second 4773 the server became unreachable and was undeclared the system peer. ten minutes later I reconnected the server which again became reachable and the system peer. You are missing the point. The bug doesn't show up if the clock filter is filled with 8 valid samples before the connection is blocked. There has to be at least one MAXDISPERSE sample (not the last one). Blocking the connection on ntpd start right after fourth reply was recevied seems to trigger the bug reliably. If you want to trigger it in a normal situation, you need to partially flush the filter, let few valid samples in and then block the connection pernamently. This doesn't work reliably and you might need to retry it several times. If you are using an old release version, that might explain your results. I can reproduce it with 4.2.6p1 and 4.2.7p32, I haven't tried anything else. -- Miroslav Lichvar ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] How should an NTP server fail?
Miroslav, Careful inspection of the code, experience with the debug trace and the evidence presented previously strongly suggests your conclusion has no relevance. See my previous messages on the number of samples in the clock filter. I have nothing further to offer you. Dave Miroslav Lichvar wrote: On Fri, Jun 11, 2010 at 03:08:53AM +, David L. Mills wrote: An hour later I disconnected the server. By second 4773 the server became unreachable and was undeclared the system peer. ten minutes later I reconnected the server which again became reachable and the system peer. You are missing the point. The bug doesn't show up if the clock filter is filled with 8 valid samples before the connection is blocked. There has to be at least one MAXDISPERSE sample (not the last one). Blocking the connection on ntpd start right after fourth reply was recevied seems to trigger the bug reliably. If you want to trigger it in a normal situation, you need to partially flush the filter, let few valid samples in and then block the connection pernamently. This doesn't work reliably and you might need to retry it several times. If you are using an old release version, that might explain your results. I can reproduce it with 4.2.6p1 and 4.2.7p32, I haven't tried anything else. ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] How should an NTP server fail?
Miroslav, My scenario did include the condition of eight valid sample before pulling the plug The first missed poll response did result in one MAXDISPERSE sample in the filter, but that is completely irrelevant. The point is the poll routine calls clock_select() for every poll sent when the last four samples in the reach register are missing, so it is guaranteed that the server be marked down after that. You are invited to peruse the code in ntp_proto.c and look for calls to clock_select(). Note it is not possible to exit this routine without either finding a system peer or declaring no system peer. See the calls to report_event(). Dave Miroslav Lichvar wrote: On Fri, Jun 11, 2010 at 03:08:53AM +, David L. Mills wrote: An hour later I disconnected the server. By second 4773 the server became unreachable and was undeclared the system peer. ten minutes later I reconnected the server which again became reachable and the system peer. You are missing the point. The bug doesn't show up if the clock filter is filled with 8 valid samples before the connection is blocked. There has to be at least one MAXDISPERSE sample (not the last one). Blocking the connection on ntpd start right after fourth reply was recevied seems to trigger the bug reliably. If you want to trigger it in a normal situation, you need to partially flush the filter, let few valid samples in and then block the connection pernamently. This doesn't work reliably and you might need to retry it several times. If you are using an old release version, that might explain your results. I can reproduce it with 4.2.6p1 and 4.2.7p32, I haven't tried anything else. ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] How should an NTP server fail?
On 2010-06-05, goofyzig backup...@gmail.com wrote: Hi there. A quick question about what to expect when it comes to NTP failures. I configured a Meinberg NTP server (software-based server), had it working, serving time to the back-end hosts, Reachability at 377 for its two NTP upstream time sources (internet NTP servers). Once configured, I wanted to see how the server failed. Please post the configuration file you used as the version line from 'ntpq -crv' (on the server). -- Steve Kostecke koste...@ntp.org NTP Public Services Project - http://support.ntp.org/ ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] How should an NTP server fail?
On 2010-06-11, Kasper Pedersen s...@kasperkp.dk wrote: On 06/11/2010 04:47 PM, David L. Mills wrote: My scenario did include the condition of eight valid sample before pulling the plug The first missed poll response did result in one MAXDISPERSE sample in the filter, but that is completely irrelevant. The point is the poll routine calls clock_select() for every poll sent when the last four samples in the reach register are missing, so it is guaranteed that the server be marked down after that. For entertainment value, 'ntpq cartoons' with 4 samples, and 8 samples: At 2 minutes after pulling the plug, there is a difference. I can confirm the observed behavior with 4.2.7p34. However the inflection point is actually between the 4th and 5th polls. My experiments this afternoon also showed that the presence or absence of iburst on the server line has no effect. -- Steve Kostecke koste...@ntp.org NTP Public Services Project - http://support.ntp.org/ ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
[ntp:questions] proprietary hardware clock as NTP reference source
Hello list, I have a set of proprietary hardware timing cards (Symmetricom bc635PCIe) which synchronize their clocks using a dedicated interconnect. As you might imagine the timing card conditioned time drifts from that of the hosts they're installed in. What I'd like to do is make the master timing card's time into a NTP reference clock then use NTP to distribute that time to the other hosts in the (private) network. I've looked at Orphan mode and undisciplined local clocks but they only refer to the host's software clock if I'm not mistaken. I've also searched through the archive but I'm afraid I lack the appropriate terminology to get meaningful results. Can someone point me toward (some google words maybe) creating an arbitrary NTP reference source (under Linux)? I think I'm just missing something very basic. I've been looking at the LinuxPPS project as these cards output a PPS that I might use to condition the host clock using Linux PPS but I don't have a 8250 serial port on these new fangled PCs. TIA! Andy ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] proprietary hardware clock as NTP reference source
In article 40aa2d27-9826-4948-a405-e5b311358...@z8g2000yqz.googlegroups.com, apobrien apobriens...@gmail.com writes: Hello list, I have a set of proprietary hardware timing cards (Symmetricom bc635PCIe) which synchronize their clocks using a dedicated interconnect. As you might imagine the timing card conditioned time drifts from that of the hosts they're installed in. What I'd like to do is make the master timing card's time into a NTP reference clock then use NTP to distribute that time to the other hosts in the (private) network. I've looked at Orphan mode and undisciplined local clocks but they only refer to the host's software clock if I'm not mistaken. I've also searched through the archive but I'm afraid I lack the appropriate terminology to get meaningful results. Can someone point me toward (some google words maybe) creating an arbitrary NTP reference source (under Linux)? I think I'm just missing something very basic. I've been looking at the LinuxPPS project as these cards output a PPS that I might use to condition the host clock using Linux PPS but I don't have a 8250 serial port on these new fangled PCs. TIA! Andy I think you have two choices. One is to write a stand alone program that talks to your hardware and puts the info into shared memory where the shared-memory driver (driver 28) can get it. gpsd works this way. The other is to write your own refclock-driver. The usual approach is to find a driver that is as close as you can get and modify it. If your changes are small and don't break the old moce, you may be able to convince the ntp project to merge them into the main source package. It may be tricky to get started this way. The best documentation is to read the code. There is some overview here, but it probably won't help much until you look at the code: http://www.cis.udel.edu/~mills/ntp/html/howto.html -- These are my opinions, not necessarily my employer's. I hate spam. ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] proprietary hardware clock as NTP reference source
On 2010-06-11, apobrien apobriens...@gmail.com wrote: Hello list, I have a set of proprietary hardware timing cards (Symmetricom bc635PCIe) which synchronize their clocks using a dedicated interconnect. As you might imagine the timing card conditioned time drifts from that of the hosts they're installed in. What I'd like to do is make the master timing card's time into a NTP reference clock then use NTP to distribute that time to the other hosts in the (private) network. I've looked at Orphan mode and undisciplined local clocks but they only refer to the host's software clock if I'm not mistaken. I've also searched through the archive but I'm afraid I lack the appropriate terminology to get meaningful results. Can someone point me toward (some google words maybe) creating an arbitrary NTP reference source (under Linux)? I think I'm just missing something very basic. I've been looking at the LinuxPPS project as these cards output a PPS that I might use to condition the host clock using Linux PPS but I don't have a 8250 serial port on these new fangled PCs. Use shmpps. It uses the shm driver to input the PPS to ntp. It requires that you have some other source to derive the seconds from (I, PPS can allow ntp to condition to the exact second, but has no idea which second it is conditioning to) That serial port is used to input the PPS and the seconds. How would you input the PPS in your case? TIA! Andy ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions