[ntp:questions] Slow convergence loopstats (but nice results)

schmidt.rich at gmail.com schmidt.rich at gmail.com
Fri Nov 22 19:27:54 UTC 2013


On Friday, November 22, 2013 12:38:36 PM UTC-5, unruh wrote:
> On 2013-11-22, schmidt.rich at gmail.com <schmidt.rich at gmail.com> wrote:
> 
> > I have just written a PHC driver for NTP and tested it on this system:
> 
> > Supermicro SYS-50150EHF-D525 which has a pair of  Intel 82574L NICs which
> 
> > have IEEE 1588 hardware-based timestamping. I'm using NTP dev 4.2.7p397 on Linux kernel 3.12 with linuxptp. One of the PHCs is sync'd via PTP to an FEI Zyfer Gsync GrandMaster, which is in turn synced via 5MHz to the USNO Master Clock #2. 
> 
> >
> 
> > I'm running ptp4l to sync PHC1 to the GrandMaster. Then NTP is reading the refclocks PHC0, PHC1 and an NTP server on the LAN ptp2:
> 
> >
> 
> > ntpq -p
> 
> >      remote           refid      st t when poll reach   delay   offset  jitter
> 
> >==============================================================================
> 
> > +PHC(0)          .PTP.            0 l   15   16  377    0.000    0.000   0.000
> 
> > *PHC(1)          .PTP.            0 l    2   16  377    0.000   -0.001   0.000
> 
> > +ptp2            .IRIG.           1 u   38   64  377    0.123    0.018   0.007
> 
> >
> 
> > After about 15 hours the loopstats shows a s.d. of +/- 0.579 microsec with peak-peak 2.52 microsec  (3,073 points).  Very superb. 
> 
> >
> 
> > However, it took fully 75 minutes at start to converge.  It took that long to remove 20ms of phase error.  I have never seen such a slow convergence. Very smooth too.  I have tested the NMEA/ATOM drivers on this system and the convergence was the normal few minutes.  Any suggestions? Can email plots.
> 
> 
> 
> David Mills has always stated that rate of convergence is not a problem
> 
> he cares about at all. Stability of running is. ntpd is NOT designed to
> 
> converge quickly. If you want faster convergence ( and better clock
> 
> discipline although possibly slightly higher drift noise) use chrony. 
> 
> Certainly a SD of .5us is far better than any standard PC can get from
> 
> an interrupt driven stratum 0 source due to
> 
> interrupt latency fluctuations. I am surprized you can get such good
> 
> results from a NIC based solution (even with hardware timestamping).
> 
>  
> 
> >
> 
> > Rich Schmidt
> 
> > Time Service Dept
> 
> > US Naval Observatory

Understood. Yes, just curious. Currently my loopstats looks like this:

56618 69198.256 0.000000157 -4.038 0.000000483 0.000045 4
56618 69214.256 0.000000141 -4.038 0.000000482 0.000044 4
56618 69230.255 0.000000123 -4.038 0.000000482 0.000042 4
56618 69246.255 0.000000137 -4.038 0.000000481 0.000041 4
56618 69262.256 0.000000271 -4.038 0.000000481 0.000047 4
56618 69278.255 -0.000000120 -4.038 0.000000480 0.000045 4
56618 69294.255 0.000000063 -4.038 0.000000480 0.000043 4
56618 69310.255 -0.000000200 -4.038 0.000000479 0.000045 4
56618 69326.255 0.000000045 -4.038 0.000000479 0.000043 4
56618 69342.255 0.000000018 -4.038 0.000000479 0.000040 4
56618 69358.256 0.000000052 -4.038 0.000000479 0.000038 4
56618 69374.256 -0.000000409 -4.038 0.000000478 0.000048 4
56618 69390.256 -0.000000176 -4.038 0.000000478 0.000048 4
56618 69406.256 -0.000000826 -4.038 0.000000503 0.000083 4
56618 69422.256 -0.000000668 -4.038 0.000000500 0.000098 4
56618 69438.255 -0.000001028 -4.039 0.000000497 0.000126 4
56618 69454.256 -0.000000961 -4.039 0.000000494 0.000146 4
56618 69470.256 -0.000000898 -4.039 0.000000492 0.000156 4
56618 69486.255 -0.000000908 -4.039 0.000000490 0.000167 4
56618 69502.255 -0.000000916 -4.040 0.000000489 0.000173 4
56618 69518.256 -0.000000660 -4.040 0.000000487 0.000173 4
56618 69534.255 -0.000000817 -4.040 0.000000486 0.000176 4
56618 69550.255 -0.000000319 -4.040 0.000000488 0.000167 4
56618 69566.254 -0.000000368 -4.040 0.000000486 0.000159 4
56618 69582.255 -0.000000257 -4.040 0.000000485 0.000151 4
56618 69598.256 -0.000000152 -4.040 0.000000484 0.000142 4
56618 69614.255 -0.000000218 -4.040 0.000000483 0.000134 4
56618 69630.256 0.000000032 -4.040 0.000000482 0.000125 4
56618 69646.255 -0.000000193 -4.040 0.000000482 0.000118 4
56618 69662.255 -0.000000076 -4.040 0.000000481 0.000111 4
56618 69678.255 0.000000197 -4.040 0.000000481 0.000105 4
56618 69694.255 0.000000438 -4.040 0.000000480 0.000105 4
56618 69710.256 0.000000031 -4.040 0.000000480 0.000098 4
56618 69726.256 0.000000220 -4.040 0.000000479 0.000093 4
56618 69742.254 0.000000779 -4.040 0.000000490 0.000112 4
56618 69758.255 0.000000721 -4.040 0.000000488 0.000120 4
56618 69774.256 0.000000349 -4.040 0.000000487 0.000117 4
56618 69790.255 0.000000612 -4.039 0.000000486 0.000122 4
56618 69806.255 0.000000637 -4.039 0.000000485 0.000126 4
56618 69822.255 0.000000353 -4.039 0.000000484 0.000123 4
56618 69838.256 0.000000660 -4.039 0.000000483 0.000127 4
56618 69854.255 0.000000179 -4.039 0.000000483 0.000120 4
56618 69870.256 0.000000327 -4.039 0.000000482 0.000115 4
56618 69886.256 0.000000252 -4.039 0.000000481 0.000110 4
56618 69902.256 0.000000306 -4.039 0.000000481 0.000106 4
56618 69918.256 0.000000102 -4.039 0.000000480 0.000100 4
56618 69934.256 0.000000309 -4.039 0.000000480 0.000097 4
56618 69950.254 0.000000384 -4.039 0.000000479 0.000097 4
56618 69966.256 0.000000161 -4.039 0.000000479 0.000092 4
56618 69982.256 0.000000026 -4.039 0.000000479 0.000086 4



More information about the questions mailing list