flipping between stratum 1 clocks regularily - NTP

This is a discussion on flipping between stratum 1 clocks regularily - NTP ; Hello, I've recently been provided with 4 "highly accurate" ntp servers that are using GPS as their primary synchronization source. Well, 2 of the hosts are using GPS (stratum 1) and the other two are syncing to those (stratum 2). ...

+ Reply to Thread
Results 1 to 5 of 5

Thread: flipping between stratum 1 clocks regularily

  1. flipping between stratum 1 clocks regularily

    Hello,

    I've recently been provided with 4 "highly accurate" ntp servers that
    are using GPS as their primary synchronization source. Well, 2 of the
    hosts are using GPS (stratum 1) and the other two are syncing to those
    (stratum 2). I've been going around configuring all 4 hosts in my
    ntp.conf and am starting to notice that my machines are flipping back
    and forth between the two stratum 1s. I don't think this is
    desirable.

    Some of the documentation from the clock vendors suggests that their
    clocks are "so accurate" that ntp has a hard time determining which is
    more so and this affect can occur. I was thinking of dropping one of
    the ST1 clocks from the config and going with 1 ST1 and 2 ST2s....
    that seems more stable in a test. I wanted to see if anyone else has
    dealt with something similar and how you fixed it.

    Thanks.

  2. Re: flipping between stratum 1 clocks regularily

    You can try using the "prefer" flag (read the docs) to have it stick to one
    of the servers.

    Jason


    > I've recently been provided with 4 "highly accurate" ntp servers that
    > are using GPS as their primary synchronization source. Well, 2 of the
    > hosts are using GPS (stratum 1) and the other two are syncing to those
    > (stratum 2). I've been going around configuring all 4 hosts in my
    > ntp.conf and am starting to notice that my machines are flipping back
    > and forth between the two stratum 1s. I don't think this is
    > desirable.
    >
    > Some of the documentation from the clock vendors suggests that their
    > clocks are "so accurate" that ntp has a hard time determining which is
    > more so and this affect can occur. I was thinking of dropping one of
    > the ST1 clocks from the config and going with 1 ST1 and 2 ST2s....
    > that seems more stable in a test. I wanted to see if anyone else has
    > dealt with something similar and how you fixed it.


  3. Re: flipping between stratum 1 clocks regularily

    On Feb 19, 11:33 am, ja...@extremeoverclocking.com (Jason Rabel)
    wrote:
    > You can try using the "prefer" flag (read the docs) to have it stick to one
    > of the servers.
    >
    > Jason
    >
    > > I've recently been provided with 4 "highly accurate" ntp servers that
    > > are using GPS as their primary synchronization source. Well, 2 of the
    > > hosts are using GPS (stratum 1) and the other two are syncing to those
    > > (stratum 2). I've been going around configuring all 4 hosts in my
    > > ntp.conf and am starting to notice that my machines are flipping back
    > > and forth between the two stratum 1s. I don't think this is
    > > desirable.

    >
    > > Some of the documentation from the clock vendors suggests that their
    > > clocks are "so accurate" that ntp has a hard time determining which is
    > > more so and this affect can occur. I was thinking of dropping one of
    > > the ST1 clocks from the config and going with 1 ST1 and 2 ST2s....
    > > that seems more stable in a test. I wanted to see if anyone else has
    > > dealt with something similar and how you fixed it.


    Thanks Jason. I'll try that.

  4. Re: flipping between stratum 1 clocks regularily

    "mmoore0011@hotmail.com" writes:

    >Hello,


    >I've recently been provided with 4 "highly accurate" ntp servers that
    >are using GPS as their primary synchronization source. Well, 2 of the
    >hosts are using GPS (stratum 1) and the other two are syncing to those
    >(stratum 2). I've been going around configuring all 4 hosts in my
    >ntp.conf and am starting to notice that my machines are flipping back
    >and forth between the two stratum 1s. I don't think this is
    >desirable.


    A The accuracy is not limited by teh clock. It is limited by the ability of
    the computer to respond to the interrupt and read the system time. That is
    about 3usec. Even if the clock signal is better than 1usec, the computer
    cannot achieve that accuracy.


    >Some of the documentation from the clock vendors suggests that their
    >clocks are "so accurate" that ntp has a hard time determining which is
    >more so and this affect can occur. I was thinking of dropping one of
    >the ST1 clocks from the config and going with 1 ST1 and 2 ST2s....
    >that seems more stable in a test. I wanted to see if anyone else has
    >dealt with something similar and how you fixed it.


    So make one the preferred and the other a backup.


    >Thanks.


  5. Re: flipping between stratum 1 clocks regularily

    mmoore0011@hotmail.com wrote:
    > Hello,
    >
    > I've recently been provided with 4 "highly accurate" ntp servers that
    > are using GPS as their primary synchronization source. Well, 2 of the
    > hosts are using GPS (stratum 1) and the other two are syncing to those
    > (stratum 2). I've been going around configuring all 4 hosts in my
    > ntp.conf and am starting to notice that my machines are flipping back
    > and forth between the two stratum 1s. I don't think this is
    > desirable.


    Do they always have *'s or +'s on the ntpq peers display. If so, I
    believe that they will be applying exactly the same time correction as
    if they always had the same pattern of *'s and +'s. That's because the
    time correction is a weighted average of all those time sources, which
    are so marked.

    The hopping only affects the reference identity and certain quality
    metrics that are passed to clients.

+ Reply to Thread