ntp clients out of sync with server - NTP

This is a discussion on ntp clients out of sync with server - NTP ; Hello, I have a test environment, Server(WinXP2,Meinberg ntp-4.2.4p5) | __________________________________________________ __ | | | C1(XP2,ntp-4.2.4p5) C2(Vista-64,ntp-4.2.4p5) C3(CentOS5.2,ntp-4.2.2p1) Since it's time island, no connection to public time server. I select the orphan mode. After the server and clients start up. At ...

+ Reply to Thread
Results 1 to 16 of 16

Thread: ntp clients out of sync with server

  1. ntp clients out of sync with server

    Hello,

    I have a test environment,
    Server(WinXP2,Meinberg ntp-4.2.4p5)
    |
    __________________________________________________ __
    | | |
    C1(XP2,ntp-4.2.4p5) C2(Vista-64,ntp-4.2.4p5)
    C3(CentOS5.2,ntp-4.2.2p1)

    Since it's time island, no connection to public time server. I select
    the orphan mode. After the server and clients start up. At first, the
    clents can sync to server's clock. But after 1 day, all 3 clients
    begin to sync to its local clock, not the server. The output of ntpq -
    p is,
    remote refid st t when poll reach delay offset jitter
    ================================================== ==========
    10.200.98.51 127.0.0.1 5 u 105 256 377
    0.299 -1.886 0.591
    *LOCAL(0) .LOCL. 12 l 23 64 377 0.000
    0.000 0.001

    The ntp.conf on server is,
    driftfile "C:\Program Files\NTP\etc\ntp.drift"
    tos orphan 5

    The ntp.conf on all 3 clients is(except the drift file location),

    driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
    server 10.200.98.51 iburst

    server 127.127.1.0
    fudge 127.127.1.0 stratum 12
    logconfig = all
    logfile "C:\Program Files (x86)\NTP\etc\ntp.log"
    statsdir "C:\Program Files (x86)\NTP\etc\log\"
    statistics loopstats peerstats clockstats
    filegen loopstats file loopstats type day enable
    filegen peerstats file peerstats type day enable
    filegen clockstats file clockstats type day enable

    At beginning, the configuration file on client has only first 2 lines,
    the lines from server 127.127.1.0 were added for debugging purpose.
    The same thing happened two years ago, on different computer hardware.
    At that time,I select undiscipline clock on server, and want 5 clients
    sync to server. But after days running, the clients out of sync with
    server, chasing their own clock. So I have to disable ntpd, and add
    ntpdate to crond to sync the time by force.

    Could any one give me some clues on this issue. Thanks

  2. Re: ntp clients out of sync with server

    "naihong@gmail.com" writes:

    >Hello,


    >I have a test environment,
    > Server(WinXP2,Meinberg ntp-4.2.4p5)
    > |
    > __________________________________________________ __
    > | | |
    >C1(XP2,ntp-4.2.4p5) C2(Vista-64,ntp-4.2.4p5)
    >C3(CentOS5.2,ntp-4.2.2p1)


    >Since it's time island, no connection to public time server. I select
    >the orphan mode. After the server and clients start up. At first, the
    >clents can sync to server's clock. But after 1 day, all 3 clients
    >begin to sync to its local clock, not the server. The output of ntpq -
    >p is,
    > remote refid st t when poll reach delay offset jitter
    > ================================================== ==========
    > 10.200.98.51 127.0.0.1 5 u 105 256 377
    >0.299 -1.886 0.591
    > *LOCAL(0) .LOCL. 12 l 23 64 377 0.000
    >0.000 0.001


    >The ntp.conf on server is,
    >driftfile "C:\Program Files\NTP\etc\ntp.drift"
    >tos orphan 5


    >The ntp.conf on all 3 clients is(except the drift file location),


    >driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
    >server 10.200.98.51 iburst


    >server 127.127.1.0
    >fudge 127.127.1.0 stratum 12
    >logconfig = all
    >logfile "C:\Program Files (x86)\NTP\etc\ntp.log"
    >statsdir "C:\Program Files (x86)\NTP\etc\log\"
    >statistics loopstats peerstats clockstats
    >filegen loopstats file loopstats type day enable
    >filegen peerstats file peerstats type day enable
    >filegen clockstats file clockstats type day enable


    >At beginning, the configuration file on client has only first 2 lines,
    >the lines from server 127.127.1.0 were added for debugging purpose.


    Debugging purposes? What debugging purposes?

    >The same thing happened two years ago, on different computer hardware.
    >At that time,I select undiscipline clock on server, and want 5 clients
    >sync to server. But after days running, the clients out of sync with
    >server, chasing their own clock. So I have to disable ntpd, and add
    >ntpdate to crond to sync the time by force.


    And you never thought of disabling the local clock? Why?


    >Could any one give me some clues on this issue. Thanks


    You stick the local clock in why? "doctor it hurts when I beat my head
    against the wall" .
    Look at the peerstats log-- are they getting any measurements from the
    proper host?

    What is the output of ntpq -p?


  3. Re: ntp clients out of sync with server

    On 2008-10-28, naihong@gmail.com wrote:

    > ... time island ...


    [snip]

    > I select the orphan mode. After the server and clients start up. At
    > first, the clents can sync to server's clock. But after 1 day, all 3
    > clients begin to sync to its local clock, not the server.
    >
    > The output of ntpq -p is,
    >
    > remote refid st t when poll reach delay offset jitter
    > ================================================== =============
    > 10.200.98.51 127.0.0.1 5 u 105 256 377 0.299 -1.886 0.591
    > *LOCAL(0) .LOCL. 12 l 23 64 377 0.000 0.000 0.001
    >
    > The ntp.conf on server is,
    > driftfile "C:\Program Files\NTP\etc\ntp.drift"
    > tos orphan 5
    >
    > The ntp.conf on all 3 clients is(except the drift file location),
    >
    > driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
    > server 10.200.98.51 iburst
    > server 127.127.1.0
    > fudge 127.127.1.0 stratum 12


    The "clients" should not be using the Undisciplined Local Clock.

    > logconfig = all
    > logfile "C:\Program Files (x86)\NTP\etc\ntp.log"
    > statsdir "C:\Program Files (x86)\NTP\etc\log\"
    > statistics loopstats peerstats clockstats
    > filegen loopstats file loopstats type day enable
    > filegen peerstats file peerstats type day enable
    > filegen clockstats file clockstats type day enable
    >
    > At beginning, the configuration file on client has only first 2 lines,
    > the lines from server 127.127.1.0 were added for debugging purpose.


    That is a misuse of the Undisciplined Local Clock.

    > The same thing happened two years ago, on different computer hardware.
    > At that time,I select undiscipline clock on server, and want 5 clients
    > sync to server. But after days running, the clients out of sync with
    > server, chasing their own clock. So I have to disable ntpd, and add
    > ntpdate to crond to sync the time by force.
    >
    > Could any one give me some clues on this issue.


    Don't use the Undisciplined Local Clock on any ntpd which does not serve
    time to others.

    --
    Steve Kostecke
    NTP Public Services Project - http://support.ntp.org/

  4. Re: ntp clients out of sync with server

    I know 127.127.1.0 should appear in any clients' configure file. But
    without this statement, the display from ntpq -p, only display the
    client out of sync with server. I mean tally '*' is missing from
    server, I did not know which time the client followed. So add
    127.127.1.0, just make sure the client sync to its own clock. At
    beginning, I just follow previous posts, use the simplest
    configuration for orphan mode

    server's ntp.conf
    driftfile "C:\Program Files\NTP\etc\ntp.drift"
    tos orphan 10

    client's ntp.conf
    driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
    server 10.200.98.51 iburst

    But this doesn't work. The clients cannot follow server's time after
    about one day running.

    On Oct 28, 12:33*pm, Steve Kostecke wrote:
    > On 2008-10-28, naih...@gmail.com wrote:
    >
    > > ... time island ...

    >
    > [snip]
    >
    >
    >
    > > I select the orphan mode. After the server and clients start up. At
    > > first, the clents can sync to server's clock. But after 1 day, all 3
    > > clients begin to sync to its local clock, not the server.

    >
    > > The output of ntpq -p is,

    >
    > > remote * * * *refid * *st t when poll reach delay offset jitter
    > > ================================================== =============
    > > 10.200.98.51 * * * 127.0.0.1 5 u *105 *256 *377 *0.299 -1..886 *0.591
    > > *LOCAL(0) * * .LOCL. * 12 l * 23 * 64 *377 *0.000 *0.000 *0.001

    >
    > > The ntp.conf on server is,
    > > driftfile "C:\Program Files\NTP\etc\ntp.drift"
    > > tos orphan 5

    >
    > > The ntp.conf on all 3 clients is(except the drift file location),

    >
    > > driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
    > > server 10.200.98.51 iburst
    > > server 127.127.1.0
    > > fudge 127.127.1.0 stratum 12

    >
    > The "clients" should not be using the Undisciplined Local Clock.
    >
    > > logconfig = all
    > > logfile "C:\Program Files (x86)\NTP\etc\ntp.log"
    > > statsdir "C:\Program Files (x86)\NTP\etc\log\"
    > > statistics loopstats peerstats clockstats
    > > filegen loopstats file loopstats type day enable
    > > filegen peerstats file peerstats type day enable
    > > filegen clockstats file clockstats type day enable

    >
    > > At beginning, the configuration file on client has only first 2 lines,
    > > the lines from server 127.127.1.0 were added for debugging purpose.

    >
    > That is a misuse of the Undisciplined Local Clock.
    >
    > > The same thing happened two years ago, on different computer hardware.
    > > At that time,I select undiscipline clock on server, and want 5 clients
    > > sync to server. But after days running, the clients out of sync with
    > > server, chasing their own clock. So I have to disable ntpd, and add
    > > ntpdate to crond to sync the time by force.

    >
    > > Could any one give me some clues on this issue.

    >
    > Don't use the Undisciplined Local Clock on any ntpd which does not serve
    > time to others.
    >
    > --
    > Steve Kostecke
    > NTP Public Services Project -http://support.ntp.org/



  5. Re: ntp clients out of sync with server

    naihong@gmail.com wrote:
    > I know 127.127.1.0 should appear in any clients' configure file. But
    > without this statement, the display from ntpq -p, only display the
    > client out of sync with server. I mean tally '*' is missing from


    In that case the server is being treated as invalid. You will need to
    provide the output of the ntpq rv command, run against the association
    id (assoc subcommand) of the server.

    If you have any w32time servers upstream of you, please note that
    w32time doesn't honour the maximum distance heuristic, so will report a
    low stratum number even when the root dispersion is impossibly high,
    because the server hasn't been synchronised for many days, e.g we have
    one in the office that is reporting stratum three, even though it has
    not been synchronised for so long that the time is almost 7 seconds out.

    > server, I did not know which time the client followed. So add
    > 127.127.1.0, just make sure the client sync to its own clock. At


    Clients are ALWAYS synchronised to their own clock (for NTPv4)!

  6. Re: ntp clients out of sync with server

    On Oct 28, 12:21*pm, Unruh wrote:
    > "naih...@gmail.com" writes:
    > >Hello,
    > >I have a test environment,
    > > * *Server(WinXP2,Meinberg ntp-4.2.4p5)
    > > * * * * * * * * * * * * * * * * * ** * * * *|
    > > *_________________________________________________ ___
    > > *| * * * * * * * * * * * * * * * * * * * * * | * * * * * * * * * * * * * * * * * * * * |
    > >C1(XP2,ntp-4.2.4p5) *C2(Vista-64,ntp-4.2.4p5)
    > >C3(CentOS5.2,ntp-4.2.2p1)
    > >Since it's time island, no connection to public time server. I select
    > >the orphan mode. After the server and clients start up. At first, the
    > >clents can sync to server's clock. But after 1 day, all 3 clients
    > >begin to sync to its local clock, not the server. The output of ntpq -
    > >p is,
    > > * *remote * * * * *refid * * * * * st *t whenpoll *reach * *delay * *offset * * jitter
    > > * * * ================================================== ==========
    > > * * * 10.200.98.51 127.0.0.1 * * * * * *5 u * 105 *256 * 377
    > >0.299 * -1.886 * * 0.591
    > > * * * **LOCAL(0) * * * * * .LOCL. * * * * *12 l * * *23 * 64 * 377 * * 0.000
    > >0.000 * *0.001
    > >The ntp.conf on server is,
    > >driftfile "C:\Program Files\NTP\etc\ntp.drift"
    > >tos orphan 5
    > >The ntp.conf on all 3 clients is(except the drift file location),
    > >driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
    > >server 10.200.98.51 iburst
    > >server 127.127.1.0
    > >fudge 127.127.1.0 stratum 12
    > >logconfig = all
    > >logfile "C:\Program Files (x86)\NTP\etc\ntp.log"
    > >statsdir "C:\Program Files (x86)\NTP\etc\log\"
    > >statistics loopstats peerstats clockstats
    > >filegen loopstats file loopstats type day enable
    > >filegen peerstats file peerstats type day enable
    > >filegen clockstats file clockstats type day enable
    > >At beginning, the configuration file on client has only first 2 lines,
    > >the lines from server 127.127.1.0 were added for debugging purpose.

    >
    > Debugging purposes? What debugging purposes?
    >

    I mean from these logs, I want to find out why clients are out of sync
    with server. I don't think it is cause by server's local clock
    drifting, since even I exchange server and client, out of sync still
    happens.

    > >The same thing happened two years ago, on different computer hardware.
    > >At that time,I select undiscipline clock on server, and want 5 clients
    > >sync to server. But after days running, the clients out of sync with
    > >server, chasing their own clock. So I have to disable ntpd, and add
    > >ntpdate to crond to sync the time by force.

    >
    > And you never thought of disabling the local clock? Why?

    Do you mean remove "server 127.127.1.0" from client's configuration,
    it is not in the ntp.conf at beginning. Add it just for tracing
    problem purpose.


    >
    > >Could any one give me some clues on this issue. Thanks

    >
    > You stick the local clock in why? "doctor it hurts when I beat my head
    > against the wall" .
    > Look at the peerstats log-- are they getting any measurements from the
    > proper host?
    >
    > What is the output of ntpq -p?

    As previous message
    remote refid st t when poll reach delay
    offset jitter
    ================================================== ==========
    10.200.98.51 127.0.0.1 5 u 105 256 377 0.299 -1.886
    0.591
    *LOCAL(0) .LOCL. 12 l 23 64 377 0.000 0.000
    0.001


  7. Re: ntp clients out of sync with server

    "naihong@gmail.com" writes:

    >I know 127.127.1.0 should appear in any clients' configure file. But


    That should read "...should NEVER appear".

    >without this statement, the display from ntpq -p, only display the
    >client out of sync with server. I mean tally '*' is missing from


    None.

    >server, I did not know which time the client followed. So add
    >127.127.1.0, just make sure the client sync to its own clock. At
    >beginning, I just follow previous posts, use the simplest
    >configuration for orphan mode


    Why in the world would you want it to sync to its own clock?
    If it is not syncing to the server, THAT is the problem that needs to be
    solved, not some problem that there is no * in ntpq.

    What comes out of peerstats once you get rid of the local clock?



    >server's ntp.conf
    >driftfile "C:\Program Files\NTP\etc\ntp.drift"
    >tos orphan 10


    >client's ntp.conf
    >driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
    >server 10.200.98.51 iburst


    >But this doesn't work. The clients cannot follow server's time after
    >about one day running.


    What kind of clients are they? If ntp finds the client is too far off the
    server, it gives up and does not try anymore.

    So, start up ntpd -g with the server, and then look at the peerstats file
    to see what is happening to the offset. Then look at the loopstats file as
    well.
    Also tell us which operating system you are using. And which version of
    ntpd




    >On Oct 28, 12:33=A0pm, Steve Kostecke wrote:
    >> On 2008-10-28, naih...@gmail.com wrote:
    >>
    >> > ... time island ...

    >>
    >> [snip]
    >>
    >>
    >>
    >> > I select the orphan mode. After the server and clients start up. At
    >> > first, the clents can sync to server's clock. But after 1 day, all 3
    >> > clients begin to sync to its local clock, not the server.

    >>
    >> > The output of ntpq -p is,

    >>
    >> > remote =A0 =A0 =A0 =A0refid =A0 =A0st t when poll reach delay offset ji=

    >tter
    >> > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3 D=3D=3D=3D=3D=3D=3D=

    >=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3 D=3D=3D=3D=3D=3D=3D=3D=3D=
    >=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
    >> > 10.200.98.51 =A0 =A0 =A0 127.0.0.1 5 u =A0105 =A0256 =A0377 =A00.299 -1=

    >.886 =A00.591
    >> > *LOCAL(0) =A0 =A0 .LOCL. =A0 12 l =A0 23 =A0 64 =A0377 =A00.000 =A00.00=

    >0 =A00.001
    >>
    >> > The ntp.conf on server is,
    >> > driftfile "C:\Program Files\NTP\etc\ntp.drift"
    >> > tos orphan 5

    >>
    >> > The ntp.conf on all 3 clients is(except the drift file location),

    >>
    >> > driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
    >> > server 10.200.98.51 iburst
    >> > server 127.127.1.0
    >> > fudge 127.127.1.0 stratum 12

    >>
    >> The "clients" should not be using the Undisciplined Local Clock.
    >>
    >> > logconfig =3D all
    >> > logfile "C:\Program Files (x86)\NTP\etc\ntp.log"
    >> > statsdir "C:\Program Files (x86)\NTP\etc\log\"
    >> > statistics loopstats peerstats clockstats
    >> > filegen loopstats file loopstats type day enable
    >> > filegen peerstats file peerstats type day enable
    >> > filegen clockstats file clockstats type day enable

    >>
    >> > At beginning, the configuration file on client has only first 2 lines,
    >> > the lines from server 127.127.1.0 were added for debugging purpose.

    >>
    >> That is a misuse of the Undisciplined Local Clock.
    >>
    >> > The same thing happened two years ago, on different computer hardware.
    >> > At that time,I select undiscipline clock on server, and want 5 clients
    >> > sync to server. But after days running, the clients out of sync with
    >> > server, chasing their own clock. So I have to disable ntpd, and add
    >> > ntpdate to crond to sync the time by force.

    >>
    >> > Could any one give me some clues on this issue.

    >>
    >> Don't use the Undisciplined Local Clock on any ntpd which does not serve
    >> time to others.
    >>
    >> --
    >> Steve Kostecke
    >> NTP Public Services Project -http://support.ntp.org/



  8. Re: ntp clients out of sync with server

    On Oct 28, 3:03*pm, Unruh wrote:
    > "naih...@gmail.com" writes:
    > >I know 127.127.1.0 should appear in any clients' *configure file. But

    >
    > That should read "...should NEVER appear".
    >
    > >without this statement, the display from ntpq -p, *only display the
    > >client out of sync with server. I mean tally '*' is missing from

    >
    > None.
    >
    > >server, I did not know which time the client *followed. *So add
    > >127.127.1.0, just make sure the client sync to its own clock. At
    > >beginning, I just follow *previous posts, use the simplest
    > >configuration for orphan mode

    >
    > Why in the world would you want it to sync to its own clock?
    > If it is not syncing to the server, THAT is the problem that needs to be
    > solved, not some problem that there is no * in ntpq.
    >
    > What comes out of peerstats once you get rid of the local clock?
    >
    > >server's ntp.conf
    > >driftfile "C:\Program Files\NTP\etc\ntp.drift"
    > >tos orphan 10
    > >client's ntp.conf
    > >driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
    > >server 10.200.98.51 iburst
    > >But this doesn't work. The clients cannot follow server's time after
    > >about one day running.

    >
    > What kind of clients are they? If ntp finds the client is too far off the
    > server, it gives up and does not try anymore.
    >
    > So, start up ntpd -g with the server, and then look at the peerstats file
    > to see what is happening to the offset. Then look at the loopstats file as
    > well.
    > Also tell us which operating system you are using. And which version of
    > ntpd
    >

    This is a LAN environment, all computers are in the same network
    segment, same building.
    Server is WinXP XP2, install Meinberg ntpd 4.2.4p5, disable w32time.
    service ntp start up with Meinberg's default setting:
    C:\Program Files\NTP\bin\ntpd.exe -U 3 -M -g -c "C:\Program Files
    \NTP\etc\ntp.conf"

    Client1, Vista x64, install Meinberg ntpd 4.2.4p5, disable w32time
    Client2, WinXP XP3, install Meinberg ntpd 4.2.4p5, disable w32time
    Client3, CentOS 5.2, install ntp-4.2.2p1

    After get rid of local clock. One Client's peerstats:
    54768 46.406 10.200.98.51 9014 -0.038017114 0.000306409 7.937500956
    0.000000954
    54768 48.403 10.200.98.51 9014 -0.064618551 0.000291795 3.937508934
    0.026601437
    54768 50.403 10.200.98.51 9014 -0.064626478 0.000261123 1.937516673
    0.018815662
    54768 52.403 10.200.98.51 9014 -0.064552078 0.000416696 0.937522418
    0.015320077
    54768 54.402 10.200.98.51 9014 -0.064642234 0.000280164 0.437526228
    0.013312644
    54768 56.402 10.200.98.51 9014 -0.064641344 0.000289907 0.187528601
    0.011906791
    54768 58.403 10.200.98.51 9014 -0.064635652 0.000283088 0.062530022
    0.010867030
    54768 60.402 10.200.98.51 9014 -0.064647984 0.000275511 0.000030850
    0.010065599
    54768 62.402 10.200.98.51 9014 -0.064653534 0.000283184 0.000030850
    0.000042888
    54768 112.400 10.200.98.51 9014 -0.064749014 0.000277924 0.000388037
    0.000124767
    54768 178.397 10.200.98.51 9014 -0.064899955 0.000276278 0.000684287
    0.000259442
    54768 241.394 10.200.98.51 9014 -0.065027426 0.000298725 0.000808213
    0.000344021
    54768 304.391 10.200.98.51 9014 -0.065179532 0.000273336 0.000868389
    0.000452187
    54768 367.388 10.200.98.51 9014 -0.065336764 0.000279747 0.000896690
    0.000546344
    54768 430.385 10.200.98.51 9014 -0.065420572 0.000364421 0.000909053
    0.000551227
    54768 494.382 10.200.98.51 9014 -0.065626320 0.000265256 0.000920889
    0.000646756
    54768 560.379 10.200.98.51 9014 -0.065780028 0.000268319 0.000939873
    0.000666959
    54768 624.376 10.200.98.51 9014 -0.065921984 0.000273026 0.000934014
    0.000665204
    54768 690.373 10.200.98.51 9014 -0.066101032 0.000270696 0.000946025
    0.000697892
    54768 754.370 10.200.98.51 9014 -0.066225812 0.000286257 0.000937061
    0.000676072
    54768 817.367 10.200.98.51 9014 -0.066385850 0.000256096 0.000925107
    0.000686414
    54768 881.364 10.200.98.51 9014 -0.066542658 0.000294116 0.000926572
    0.000695455
    54768 945.361 10.200.98.51 9014 -0.066627817 0.000417715 0.000927277
    0.000623911
    54768 1011.358 10.200.98.51 9014 -0.066574980 0.000863147 0.000942515
    0.000454266
    54768 1075.355 10.200.98.51 9014 -0.067014922 0.000308414 0.000935247
    0.000718587
    54768 1140.352 10.200.98.51 9014 -0.066858390 0.000940082 0.000939059
    0.000455869
    54768 1206.349 10.200.98.51 9014 -0.067061963 0.000876897 0.000948435
    0.000520336
    54768 1270.346 10.200.98.51 9014 -0.067222507 0.000911194 0.000938182
    0.000552473
    54768 1335.344 10.200.98.51 9614 -0.067671481 0.000916143 0.000940468
    0.000864462
    54768 1399.336 10.200.98.51 9614 -0.067014922 0.000308414 0.005095755
    0.001572922
    54768 1463.329 10.200.98.51 9614 -0.067014922 0.000308414 0.005307104
    0.003247092
    54768 1529.322 10.200.98.51 9614 -0.056247920 0.000292353 0.002648235
    0.009650302
    54768 1595.315 10.200.98.51 9614 -0.052894009 0.000286015 0.001436868
    0.011804385
    54768 1658.309 10.200.98.51 9614 -0.052894009 0.000286015 0.001643527
    0.010620305
    54768 1723.304 10.200.98.51 9614 -0.052894009 0.000286015 0.002069093
    0.009442148

    its loopstats output:
    54768 1335.343 -0.067671481 -46.699 0.023925482 0.000000 6
    54768 1529.322 -0.056247920 -46.914 0.022741752 0.075862 6
    54768 1595.315 -0.052894009 -47.115 0.021305985 0.100622 6


    > >On Oct 28, 12:33=A0pm, Steve Kostecke wrote:
    > >> On 2008-10-28, naih...@gmail.com wrote:

    >
    > >> > ... time island ...

    >
    > >> [snip]

    >
    > >> > I select the orphan mode. After the server and clients start up. At
    > >> > first, the clents can sync to server's clock. But after 1 day, all 3
    > >> > clients begin to sync to its local clock, not the server.

    >
    > >> > The output of ntpq -p is,

    >
    > >> > remote =A0 =A0 =A0 =A0refid =A0 =A0st t when poll reach delay offset ji=

    > >tter
    > >> > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3 D=3D=3D=3D=3D=3D=3D=

    > >=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3 D=3D=3D=3D=3D=3D=3D=3D=3 D=
    > >=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
    > >> > 10.200.98.51 =A0 =A0 =A0 127.0.0.1 5 u =A0105 =A0256 =A0377 =A00.299 -1=

    > >.886 =A00.591
    > >> > *LOCAL(0) =A0 =A0 .LOCL. =A0 12 l =A0 23 =A0 64 =A0377 =A00.000 =A00.00=

    > >0 =A00.001

    >
    > >> > The ntp.conf on server is,
    > >> > driftfile "C:\Program Files\NTP\etc\ntp.drift"
    > >> > tos orphan 5

    >
    > >> > The ntp.conf on all 3 clients is(except the drift file location),

    >
    > >> > driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
    > >> > server 10.200.98.51 iburst
    > >> > server 127.127.1.0
    > >> > fudge 127.127.1.0 stratum 12

    >
    > >> The "clients" should not be using the Undisciplined Local Clock.

    >
    > >> > logconfig =3D all
    > >> > logfile "C:\Program Files (x86)\NTP\etc\ntp.log"
    > >> > statsdir "C:\Program Files (x86)\NTP\etc\log\"
    > >> > statistics loopstats peerstats clockstats
    > >> > filegen loopstats file loopstats type day enable
    > >> > filegen peerstats file peerstats type day enable
    > >> > filegen clockstats file clockstats type day enable

    >
    > >> > At beginning, the configuration file on client has only first 2 lines,
    > >> > the lines from server 127.127.1.0 were added for debugging purpose.

    >
    > >> That is a misuse of the Undisciplined Local Clock.

    >
    > >> > The same thing happened two years ago, on different computer hardware.
    > >> > At that time,I select undiscipline clock on server, and want 5 clients
    > >> > sync to server. But after days running, the clients out of sync with
    > >> > server, chasing their own clock. So I have to disable ntpd, and add
    > >> > ntpdate to crond to sync the time by force.

    >
    > >> > Could any one give me some clues on this issue.

    >
    > >> Don't use the Undisciplined Local Clock on any ntpd which does not serve
    > >> time to others.

    >
    > >> --
    > >> Steve Kostecke
    > >> NTP Public Services Project -http://support.ntp.org/



  9. Re: ntp clients out of sync with server

    On Oct 28, 2:38*pm, David Woolley
    wrote:
    > naih...@gmail.com wrote:
    > > I know 127.127.1.0 should appear in any clients' *configure file. But
    > > without this statement, the display from ntpq -p, *only display the
    > > client out of sync with server. I mean tally '*' is missing from

    >
    > In that case the server is being treated as invalid. *You will need to
    > provide the output of the ntpq rv command, run against the association
    > id (assoc subcommand) of the server.
    >

    The output from ntpq -c"rv":
    assID=0 status=0644 leap_none, sync_ntp, 4 events, event_peer/
    strat_chg,
    version="ntpd 4.2.4p5@beijing-o Sep 01 9:15:56 (UTC+02:00) 2008
    (10)",
    processor="unknown", system="WINDOWS/NT", leap=00, stratum=6,
    precision=-20, rootdelay=0.304, rootdispersion=58.322, peer=46864,
    refid=10.200.98.51,
    reftime=ccb22e3c.45338a42 Tue, Oct 28 2008 17:35:08.270, poll=6,
    clock=ccb22eb8.2242d2bb Tue, Oct 28 2008 17:37:12.133, state=4,
    offset=-33.131, frequency=-47.242, jitter=10.226, noise=21.119,
    stability=0.104, tai=0


    > If you have any w32time servers upstream of you, please note that
    > w32time doesn't honour the maximum distance heuristic, so will report a
    > low stratum number even when the root dispersion is impossibly high,
    > because the server hasn't been synchronised for many days, e.g we have
    > one in the office that is reporting stratum three, even though it has
    > not been synchronised for so long that the time is almost 7 seconds out.
    >

    w32time has been disabled on all windows based computer, no matter
    server or client. I have tested ntpd in pure linux environment before,
    both server/client are linux, and have no internet connection. It
    seems clients can find the server has untrusted time and refuse to
    sync with it after running a period of time(about 1 day).
    > > server, I did not know which time the client *followed. *So add
    > > 127.127.1.0, just make sure the client sync to its own clock. At

    >
    > Clients are ALWAYS synchronised to their own clock (for NTPv4)!



  10. Re: ntp clients out of sync with server

    naihong@gmail.com wrote:

    >>

    > The output from ntpq -c"rv":


    I asked you to do ntpq -c"rv nnnnn"
    where nnnnn is the number given by ntpq -c"assoc"
    for the relevant server.

    However, this output does indicate that you are synchronised to
    something other than the local clock. You need to run the command
    when the system is broken.

    > assID=0 status=0644 leap_none, sync_ntp, 4 events, event_peer/
    > strat_chg,
    > version="ntpd 4.2.4p5@beijing-o Sep 01 9:15:56 (UTC+02:00) 2008
    > (10)",
    > processor="unknown", system="WINDOWS/NT", leap=00, stratum=6,
    > precision=-20, rootdelay=0.304, rootdispersion=58.322, peer=46864,
    > refid=10.200.98.51,
    > reftime=ccb22e3c.45338a42 Tue, Oct 28 2008 17:35:08.270, poll=6,
    > clock=ccb22eb8.2242d2bb Tue, Oct 28 2008 17:37:12.133, state=4,
    > offset=-33.131, frequency=-47.242, jitter=10.226, noise=21.119,
    > stability=0.104, tai=0
    >
    >
    >> If you have any w32time servers upstream of you, please note that
    >> w32time doesn't honour the maximum distance heuristic, so will report a
    >> low stratum number even when the root dispersion is impossibly high,
    >> because the server hasn't been synchronised for many days, e.g we have
    >> one in the office that is reporting stratum three, even though it has
    >> not been synchronised for so long that the time is almost 7 seconds out.
    >>

    > w32time has been disabled on all windows based computer, no matter
    > server or client. I have tested ntpd in pure linux environment before,
    > both server/client are linux, and have no internet connection. It
    > seems clients can find the server has untrusted time and refuse to
    > sync with it after running a period of time(about 1 day).
    >>> server, I did not know which time the client followed. So add
    >>> 127.127.1.0, just make sure the client sync to its own clock. At

    >> Clients are ALWAYS synchronised to their own clock (for NTPv4)!

    >



  11. Re: ntp clients out of sync with server

    On Oct 29, 1:08*am, David Woolley
    wrote:
    > naih...@gmail.com wrote:
    >
    > > The output from ntpq -c"rv":

    >
    > I asked you to do ntpq -c"rv nnnnn"
    > where nnnnn is the number given by ntpq -c"assoc"
    > for the relevant server.
    >
    > However, this output does indicate that you are synchronised to
    > something other than the local clock. *You need to run the command
    > when the system is broken.
    >

    That is really help, at the time sync is successful,

    >ntpq -c"assoc"


    ind assID status conf reach auth condition last_event cnt
    ================================================== =========
    1 46864 9614 yes yes none sys.peer reachable 1

    After the sync is broken

    >ntpq -c"assoc"


    ind assID status conf reach auth condition last_event cnt
    ================================================== =========
    1 46864 9014 yes yes none reject reachable 1

    condition change from sys.peer to reject, status also changed

    >ntpq -c"rv 46864"

    assID=46864 status=9014 reach, conf, 1 event, event_reach,
    srcadr=10.200.98.51, srcport=123, dstadr=10.200.98.110, dstport=123,
    leap=00, stratum=5, precision=-20, rootdelay=0.000,
    rootdispersion=1175.797, refid=127.0.0.1, reach=377, unreach=0,
    hmode=3,
    pmode=4, hpoll=8, ppoll=8, flash=400 peer_dist, keyid=0, ttl=0,
    offset=-24.021, delay=0.294, dispersion=6.001, jitter=1.199,
    reftime=00000000.00000000 Wed, Feb 6 2036 22:28:16.000,
    org=ccb35db7.959d8613 Wed, Oct 29 2008 15:09:59.584,
    rec=ccb35db7.9bf0592f Wed, Oct 29 2008 15:09:59.609,
    xmt=ccb35db7.9bda87ad Wed, Oct 29 2008 15:09:59.608,
    filtdelay= 0.31 0.29 0.32 0.33 0.32 0.31 0.40
    0.30,
    filtoffset= -24.55 -24.02 -23.49 -23.01 -22.56 -22.15 -21.85
    -21.38,
    filtdisp= 0.00 3.84 7.71 11.54 15.41 19.28 23.10
    26.96

    If I start the ntpd client with -d option, it print out
    auth_agekeys: at 5xxx keys 1 expired 0
    looks like the sync break relate to authentication. But it is not
    enabled. Does this mean ntp4 need it in default?

    > > assID=0 status=0644 leap_none, sync_ntp, 4 events, event_peer/
    > > strat_chg,
    > > version="ntpd 4.2.4p5@beijing-o Sep 01 9:15:56 (UTC+02:00) 2008
    > > (10)",
    > > processor="unknown", system="WINDOWS/NT", leap=00, stratum=6,
    > > precision=-20, rootdelay=0.304, rootdispersion=58.322, peer=46864,
    > > refid=10.200.98.51,
    > > reftime=ccb22e3c.45338a42 *Tue, Oct 28 2008 17:35:08.270, poll=6,
    > > clock=ccb22eb8.2242d2bb *Tue, Oct 28 2008 17:37:12.133, state=4,
    > > offset=-33.131, frequency=-47.242, jitter=10.226, noise=21.119,
    > > stability=0.104, tai=0

    >
    > >> If you have any w32time servers upstream of you, please note that
    > >> w32time doesn't honour the maximum distance heuristic, so will report a
    > >> low stratum number even when the root dispersion is impossibly high,
    > >> because the server hasn't been synchronised for many days, e.g we have
    > >> one in the office that is reporting stratum three, even though it has
    > >> not been synchronised for so long that the time is almost 7 seconds out.

    >
    > > w32time has been disabled on all windows based computer, no matter
    > > server or client. I have tested ntpd in pure linux environment before,
    > > both server/client are linux, and have no internet connection. It
    > > seems clients can find the server has untrusted time and refuse to
    > > sync with it after running a period of time(about 1 day).
    > >>> server, I did not know which time the client *followed. *So add
    > >>> 127.127.1.0, just make sure the client sync to its own clock. At
    > >> Clients are ALWAYS synchronised to their own clock (for NTPv4)!



  12. Re: ntp clients out of sync with server

    naihong@gmail.com wrote:
    > condition change from sys.peer to reject, status also changed
    >
    >> ntpq -c"rv 46864"

    > assID=46864 status=9014 reach, conf, 1 event, event_reach,
    > srcadr=10.200.98.51, srcport=123, dstadr=10.200.98.110, dstport=123,
    > leap=00, stratum=5, precision=-20, rootdelay=0.000,
    > rootdispersion=1175.797, refid=127.0.0.1, reach=377, unreach=0,

    ^^^^^^^^

    rootdispersion is too high. The way that orphan mode was explained to
    me that shouldn't be possible. I believe that orphan mode should set a
    fixed, but random, value which is significantly less than 1 second.

    I wonder if it is spuriously aging the root dispersion in orphan mode.
    Or is orphan mode just being more honest (and therefore not supporting
    time islands). Would one of the orphan mode advocates like to comment?

    One side issue is that there isn't a Meinberg version of NTP, only a
    Meinberg installer for the Windows NT build of the reference implementation.

    > hmode=3,
    > pmode=4, hpoll=8, ppoll=8, flash=400 peer_dist, keyid=0, ttl=0,

    ^^^^^^^^^

    Peer dist[ance too high] is telling you the same thing.


  13. Orphan mode not suitable for time islands? (was: ntp clients outof sync with server)

    David Woolley wrote:
    > naihong@gmail.com wrote:
    >> condition change from sys.peer to reject, status also changed
    >>
    >>> ntpq -c"rv 46864"

    >> assID=46864 status=9014 reach, conf, 1 event, event_reach,
    >> srcadr=10.200.98.51, srcport=123, dstadr=10.200.98.110, dstport=123,
    >> leap=00, stratum=5, precision=-20, rootdelay=0.000,
    >> rootdispersion=1175.797, refid=127.0.0.1, reach=377, unreach=0,

    > ^^^^^^^^
    >
    > rootdispersion is too high. The way that orphan mode was explained to
    > me that shouldn't be possible. I believe that orphan mode should set a
    > fixed, but random, value which is significantly less than 1 second.
    >
    > I wonder if it is spuriously aging the root dispersion in orphan mode.
    > Or is orphan mode just being more honest (and therefore not supporting
    > time islands). Would one of the orphan mode advocates like to comment?


    Still waiting for an explanation from the orphan mode advocates. His
    server is basically an orphan with no servers and there is only one
    orphan in the whole network.

    >
    > One side issue is that there isn't a Meinberg version of NTP, only a
    > Meinberg installer for the Windows NT build of the reference
    > implementation.
    >
    >> hmode=3,
    >> pmode=4, hpoll=8, ppoll=8, flash=400 peer_dist, keyid=0, ttl=0,

    > ^^^^^^^^^
    >
    > Peer dist[ance too high] is telling you the same thing.
    >


  14. Re: ntp clients out of sync with server

    David Woolley wrote:

    > Peer dist[ance too high] is telling you the same thing.
    >

    As no-one has spoken in defence of orphan mode, I think your best course
    of action is to replace orphan mode with a high stratum local clock
    driver, on the root server of your time island.

  15. Re: ntp clients out of sync with server

    On Nov 1, 11:17*am, David Woolley
    wrote:
    > David Woolley wrote:
    > > Peer dist[ance too high] is telling you the same thing.

    >
    > As no-one has spoken in defence of orphan mode, I think your best course
    > of action is to replace orphan mode with a high stratum local clock
    > driver, on the root server of your time island.


    Thanks a lot!

  16. Re: ntp clients out of sync with server

    On 2008-11-01, David Woolley wrote:

    > David Woolley wrote:
    >
    >> Peer dist[ance too high] is telling you the same thing.

    >
    > As no-one has spoken in defence of orphan mode,


    Where on earth do you get this delusion that one of ntpd's modes of
    operation needs to be "defended"?

    Orphan Mode is discussed at length at
    http://www.cis.udel.edu/~mills/ntp/h...oc.html#orphan

    If you think that it does not work as advertised then please go ahead
    and file a bug report.

    --
    Steve Kostecke
    NTP Public Services Project - http://support.ntp.org/

+ Reply to Thread