uk pool problem - NTP

This is a discussion on uk pool problem - NTP ; For some reason, ntpd will not set my clock so I've been using ntpdate via crontab which works fine, except that occasionally it picks blackhole.arlott.org.uk which sets my clock to 23:59 on 31/12/1999. Can I blacklist this site or is ...

+ Reply to Thread
Page 1 of 4 1 2 3 ... LastLast
Results 1 to 20 of 80

Thread: uk pool problem

  1. uk pool problem

    For some reason, ntpd will not set my clock so I've been using ntpdate
    via crontab which works fine, except that occasionally it picks
    blackhole.arlott.org.uk which sets my clock to 23:59 on 31/12/1999. Can
    I blacklist this site or is there something else going on here?
    TIA,
    mike.


  2. Re: uk pool problem

    "Mike Parkins" writes:

    >For some reason, ntpd will not set my clock so I've been using ntpdate
    >via crontab which works fine, except that occasionally it picks
    >blackhole.arlott.org.uk which sets my clock to 23:59 on 31/12/1999. Can
    >I blacklist this site or is there something else going on here?


    FWIW, I see the same thing over IPv6 or IPv4 when I use ntpdate. It's
    even running a web server that returns a consistent date header. However,
    it produces moderately sane looking output to "ntpq -p".

    I guess you should contact the guys at www.pool.ntp.org?

    David.

  3. Re: uk pool problem

    You should look at:

    http://www.pool.ntp.org/scores/84.92.75.21

    We see that the server is good. If a server have a score lower than 5
    it will be removed from the pool.

    There must be another problem - that server is a good one.


  4. Re: uk pool problem

    Eugen COCA wrote:

    > http://www.pool.ntp.org/scores/84.92.75.21
    >
    > We see that the server is good. If a server have a score lower than 5
    > it will be removed from the pool.
    >
    > There must be another problem - that server is a good one.


    It wasn't about an hour ago. At least not when you threw ntpdate at it:

    thor:~$ ntpdate -q blackhole.arlott.org.uk
    server 84.92.75.21, stratum 15, offset -210105193.104270, delay 0.18675
    28 Aug 20:33:12 ntpdate[11281]: step time server 84.92.75.21 offset
    -210105193.1

    Doesn't exactly look good, does it?

    At the moment it is a) again synced to its DCF clock (it was not at the
    time of that ntpdate, although it *was* synced to one of the remote
    servers):

    thor:~$ ntpq -p blackhole.arlott.org.uk
    remote refid st t when poll reach delay offset
    jitter
    ================================================== ============================
    *SHM(0) .DCFa. 0 l 68 64 372 0.000 -2.845
    0.550
    SHM(1) .MSFa. 0 l 969 64 0 0.000 0.000
    4000.00
    xntp0.nl.uu.net .GPS. 1 u 22 1024 377 189.785 61.654
    54.753
    -ntp1.nl.uu.net .GPS. 1 u 1024 1024 375 60.115 -6.040
    0.766
    -tik.cesnet.cz .GPS. 1 u 43 1024 377 75.240 0.739
    1.891
    -ntp1.sp.se .PPS. 1 u 45 1024 377 79.036 -4.804
    2.510
    +ntp2.sp.se .PPS. 1 u 11 1024 377 87.011 -2.588
    4.270
    -MSC-101.it.nuig .GPS. 1 u 16 1024 377 85.030 1.546
    11.791
    +ntps1-0.cs.tu-b .PPS. 1 u 1016 1024 377 76.176 -3.653
    26.861
    batman.npl.co.u 139.143.49.13 2 u 926 1024 377 54.174 -3.554
    8.751
    ntp2.npl.co.uk 78.84.80.0 2 u 888 1024 377 62.029 9.998
    2.568


    and b) it does it again:

    thor:~$ ntpdate -q blackhole.arlott.org.uk
    server 84.92.75.21, stratum 15, offset -210111428.429590, delay 0.09758
    28 Aug 22:17:10 ntpdate[11984]: step time server 84.92.75.21 offset
    -210111428.429590 sec


    And 'thor' is my own stratum 1 NTP server (Motorola Oncore M12).

    The boz is *not* OK.


    Regards,

    Marco.





  5. Re: uk pool problem

    >From my Stratum 1 server:

    ntp1# ntptrace 84.92.75.21
    blackhole.arlott.org.uk: stratum 1, offset -0.002353, synch distance
    0.006055, refid 'DCFa'


    So, all things seems to be OK !


  6. Re: uk pool problem

    Eugen COCA wrote:

    >>From my Stratum 1 server:

    >
    > ntp1# ntptrace 84.92.75.21
    > blackhole.arlott.org.uk: stratum 1, offset -0.002353, synch distance
    > 0.006055, refid 'DCFa'
    >
    >
    > So, all things seems to be OK !


    One after the other:

    thor:~$ ntptrace blackhole.arlott.org.uk
    blackhole.arlott.org.uk: stratum 1, offset -0.000197, synch distance
    0.002905, refid 'DCFa'

    That's OK.

    thor:~$ ntpq -p blackhole.arlott.org.uk
    remote refid st t when poll reach delay offset
    jitter
    ================================================== ============================
    *SHM(0) .DCFa. 0 l 30 64 257 0.000 -0.197
    0.276
    SHM(1) .MSFa. 0 l 939 64 0 0.000 0.000
    4000.00
    +ntp0.nl.uu.net .GPS. 1 u 11 128 377 66.472 -0.205
    52.727
    +ntp1.nl.uu.net .GPS. 1 u 59 64 377 60.396 -0.509
    3.059
    +tik.cesnet.cz .GPS. 1 u 42 64 377 68.898 -0.375
    7.250
    +ntp1.sp.se .PPS. 1 u 33 64 377 81.693 -0.870
    19.606
    +ntp2.sp.se .PPS. 1 u 56 64 377 81.601 -0.454
    45.087
    -MSC-101.it.nuig .GPS. 1 u - 128 377 61.290 -1.355
    22.261
    -ntps1-0.cs.tu-b .PPS. 1 u 42 64 377 68.348 -6.179
    9.408
    batman.npl.co.u 139.143.49.13 2 u 152 1024 377 54.925 -1.994
    1.038
    ntp2.npl.co.uk 78.84.80.0 2 u 116 1024 377 77.089 12.996
    128.182

    So is this.

    thor:~$ ntpdate -q blackhole.arlott.org.uk
    server 84.92.75.21, stratum 15, offset -210114885.349156, delay 0.10283
    28 Aug 23:14:47 ntpdate[12137]: step time server 84.92.75.21 offset
    -210114885.349156 sec


    But this sure isn't!

    And the three commands are executed in just about the same second...

    Regards,

    Marco.


  7. Re: uk pool problem

    I'm not well informed on this ntp malarky...

    >From what others have found, do I assume that since everyone normally

    uses ntpd to keep good time, it's only people like me who have to use
    ntpdate that see this problem?

    It does seem that Mr Blackhole is misbehaving to some degree at least,
    no?


  8. Re: uk pool problem

    Mike Parkins wrote:

    >>From what others have found, do I assume that since everyone normally

    > uses ntpd to keep good time, it's only people like me who have to use
    > ntpdate that see this problem?


    That's what it looks like.


    > It does seem that Mr Blackhole is misbehaving to some degree at least,
    > no?


    Yes.

    Regards,

    Marco.



  9. Re: uk pool problem

    On 2006-08-28, M.C. van den Bovenkamp wrote:
    > Eugen COCA wrote:
    >
    >> So, all things seems to be OK !

    >
    > One after the other:




    > thor:~$ ntpdate -q blackhole.arlott.org.uk
    > server 84.92.75.21, stratum 15, offset -210114885.349156, delay 0.10283
    > 28 Aug 23:14:47 ntpdate[12137]: step time server 84.92.75.21 offset
    > -210114885.349156 sec
    >
    > But this sure isn't!


    This is an excellent example of why you need to use a sufficient number
    of time sources to rule out false tickers. And, possibly, a good
    illustration of why ntpdate is deprecated.

    'ntpd -gq' is the suggested replacment for ntpdate.

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

  10. Re: uk pool problem

    Eugen COCA wrote:

    >>From my Stratum 1 server:

    >
    > ntp1# ntptrace 84.92.75.21
    > blackhole.arlott.org.uk: stratum 1, offset -0.002353, synch distance
    > 0.006055, refid 'DCFa'
    >
    >
    > So, all things seems to be OK !
    >


    I just tried the following:
    sunblok_$ ntpdate -du blackhole.arlott.org.uk
    28 Aug 18:19:26 ntpdate[13166]: ntpdate 4.2.1p241-RC@1.1498 Wed Apr 26
    20:16:59 EDT 2006 (1)
    Looking for host blackhole.arlott.org.uk and service ntp
    host found : blackhole.arlott.org.uk
    transmit(84.92.75.21)
    receive(84.92.75.21)
    transmit(84.92.75.21)
    receive(84.92.75.21)
    transmit(84.92.75.21)
    receive(84.92.75.21)
    transmit(84.92.75.21)
    receive(84.92.75.21)
    transmit(84.92.75.21)
    server 84.92.75.21, port 123
    stratum 15, precision -20, leap 00, trust 000
    refid [84.92.75.21], delay 0.15500, dispersion 0.00000
    transmitted 4, in filter 4
    reference time: bc17c1ff.00000000 Fri, Dec 31 1999 18:59:59.000
    originate timestamp: bc17c1ff.00000000 Fri, Dec 31 1999 18:59:59.000
    transmit timestamp: c89dea6e.eef5df4c Mon, Aug 28 2006 18:19:26.933
    filter delay: 0.16431 0.15500 0.15509 0.15636
    0.00000 0.00000 0.00000 0.00000
    filter offset: -2101187 -2101187 -2101187 -2101187
    0.000000 0.000000 0.000000 0.000000
    delay 0.15500, dispersion 0.00000
    offset -210118767.738966

    28 Aug 18:19:27 ntpdate[13166]: step time server 84.92.75.21 offset
    -210118767.738966 sec

    It sure doesn't look okay to me!!!!! OTOH the output of
    sunblok_$ ntpq -p 84.92.75.21
    remote refid st t when poll reach delay offset
    jitter
    ================================================== ============================
    *SHM(0) .DCFa. 0 l 37 64 377 0.000 -1.798
    0.404
    SHM(1) .MSFa. 0 l 688 64 0 0.000 0.000
    4000.00
    +ntp0.nl.uu.net .GPS. 1 u 424 1024 377 62.042 -1.119
    1.750
    -ntp1.nl.uu.net .GPS. 1 u 418 1024 277 61.927 0.275
    1.153
    +tik.cesnet.cz .GPS. 1 u 388 1024 377 69.535 -1.671
    1.024
    -ntp1.sp.se .PPS. 1 u 386 1024 377 83.789 -2.838
    6.061
    +ntp2.sp.se .PPS. 1 u 408 1024 377 84.089 -1.783
    2.341
    +ntp-galway.hea. .GPS. 1 u 418 1024 377 69.287 -1.425
    0.733
    +ntps1-0.cs.tu-b .PPS. 1 u 406 1024 377 78.190 -2.083
    2.960
    batman.npl.co.u 139.143.49.13 2 u 62 1024 377 51.350 -3.677
    4.123
    ntp2.npl.co.uk 78.84.80.0 2 u 28 1024 377 56.097 1.949
    12.404
    sunblok_$ ntptrace 84.92.75.21
    blackhole.arlott.org.uk: stratum 1, offset -0.001798, synch distance
    0.004925, refid 'DCFa'

    Looks quite reasonable.


  11. Re: uk pool problem

    Mike,

    why do you need to use 'ntpdate" ? You may use the last version of
    ntpd.

    Regarding the pool, they have implemented a scoring algorithm by
    verifying EACH server in the pool twice per hour (the interval between
    two consecutive checks is about 25 to 35 minutes). The method they use
    for verifying is by comparing the offset obtained with the command
    "ntpq -e serveraddress" with the time reported by a server considered
    as a reference. So, if a server is bad, it will be for sure removed by
    the pool.

    So,

    1. Use ntpd not ntpdate;

    2. Use a large number of servers (more than 9);

    3. Use pool servers and manually entered servers addresses;

    Regards,


  12. Re: uk pool problem

    M.C. van den Bovenkamp wrote:
    > [...]
    > And the three commands are executed in just about the same second...


    I could think of some kind of misuse-protection, we discussed this
    several times. If a client polling rate is too high, send it the wrong
    time...

    Probably this is just an attempt of the owner of this server to reduce
    the load caused by mad clients. I just issued a sequence of ntpdate
    calls and it seems that I am blocked by that server automatically after
    a couple of requests (for a short time).

    Maybe you should contact this operator and ask him to rise his threshold
    for triggering counter measurements (in order to allow ntpdate to do its
    job).

    Best regards,
    Heiko

  13. Re: uk pool problem


    Eugen COCA wrote:
    > Mike,
    >
    > why do you need to use 'ntpdate" ? You may use the last version of
    > ntpd.


    ntpd doesn't work for me, I don't know why. Seems to run ok but never
    actually changes the clock. I'm using fr.pool.ntp.org now though, no
    problems so far..

    thanks for the advice,
    mike.


  14. Re: uk pool problem

    Heiko Gerstung wrote:

    > I could think of some kind of misuse-protection, we discussed this
    > several times. If a client polling rate is too high, send it the wrong
    > time...
    >
    > Probably this is just an attempt of the owner of this server to reduce
    > the load caused by mad clients. I just issued a sequence of ntpdate
    > calls and it seems that I am blocked by that server automatically after
    > a couple of requests (for a short time).


    This appears to be the same server described at

    http://ntp.lp0.eu/

    ntp.lp0.eu is a CNAME to blackhole.arlott.org.uk.

    Quote:

    "Access
    o This server is private use only.
    o All other users should use pool.ntp.org.
    o Anyone using ntpdate (or equivalent) or unauthorised frequent
    polling of the time will be reported to their ISP's abuse contact.
    Any queries should be directed to the hostmaster (at blackhole.arlott.org)."

    Perhaps it has crept into the pool by mistake.

    I don't see why ntpdate is quite so frowned upon.

    --
    Ronan Flood
    working for but not speaking for
    Network Services, University of London Computer Centre
    (which means: don't bother ULCC if I've said something you don't like)

  15. Re: uk pool problem


    Ronan Flood wrote:

    > "Access
    > o This server is private use only.
    > o All other users should use pool.ntp.org.



    This is nonsense !

    If the server is for private use, then why did you listed it in
    pool.ntp.org ?

    If you your server in the pool then the server is public.

    If there are ANY pool.ntp.org admin reading this thread - you have to
    find a solution to remove this kind of servers.

    The pools are a mixture of very good and very bad servers (one good
    example is the one we talked above). In the pool there are stratum 1,
    2, 3 or maybe more servers. In my opinion the admins must rethink the
    algorithm of accepting or not a server (maybe one ideea is to make
    separate pools for each stratum, I don't know).

    Anyway, the BEST ideea is to put in your ntp.conf as many servers as
    you know (9 or maybe more).


  16. Re: uk pool problem

    Eugen COCA wrote:
    > Mike,
    >
    > why do you need to use 'ntpdate" ? You may use the last version of
    > ntpd.
    >
    > Regarding the pool, they have implemented a scoring algorithm by


    > So,
    >
    > 1. Use ntpd not ntpdate;
    >
    > 2. Use a large number of servers (more than 9);


    If you need more than seven, you have chosen the wrong seven! Four
    should be sufficient for most applications.

    Four are necessary and sufficient to protect against one false-ticker
    Five are necessary and sufficient to protect against two false-tickers
    Seven are necessary and sufficient to protect against three. . . .


  17. Re: uk pool problem

    Mike Parkins wrote:

    > Eugen COCA wrote:
    >
    >>Mike,
    >>
    >>why do you need to use 'ntpdate" ? You may use the last version of
    >>ntpd.

    >
    >
    > ntpd doesn't work for me, I don't know why. Seems to run ok but never
    > actually changes the clock. I'm using fr.pool.ntp.org now though, no
    > problems so far..
    >
    > thanks for the advice,
    > mike.
    >


    Did you ask for help here??

    Post your ntp.conf file and the output of ntpq -p (after ntpd has been
    running for AT LEAST ten minutes. It works very well indeed for most of us.


  18. Re: uk pool problem

    Richard B. Gilbert wrote:
    []
    > Four are necessary and sufficient to protect against one false-ticker
    > Five are necessary and sufficient to protect against two false-tickers
    > Seven are necessary and sufficient to protect against three. . . .


    [David here]

    However, at the moment, when I use UK pool servers, more times than not
    the servers do not resolve at reboot, so I end up with no servers from the
    pool! This is the known DNS problem, which seems to have become worse
    over the last few months. Hence I need to specify other servers as well,
    and end up with more than the ideal number......

    David



  19. Re: uk pool problem

    Ronan Flood wrote:
    > Heiko Gerstung wrote:
    >
    >> I could think of some kind of misuse-protection, we discussed this
    >> several times. If a client polling rate is too high, send it the wrong
    >> time...
    >>
    >> Probably this is just an attempt of the owner of this server to reduce
    >> the load caused by mad clients. I just issued a sequence of ntpdate
    >> calls and it seems that I am blocked by that server automatically after
    >> a couple of requests (for a short time).

    >
    > This appears to be the same server described at
    >
    > http://ntp.lp0.eu/
    >
    > ntp.lp0.eu is a CNAME to blackhole.arlott.org.uk.
    >
    > Quote:
    >
    > "Access
    > o This server is private use only.
    > o All other users should use pool.ntp.org.
    > o Anyone using ntpdate (or equivalent) or unauthorised frequent
    > polling of the time will be reported to their ISP's abuse contact.
    > Any queries should be directed to the hostmaster (at blackhole.arlott.org)."
    >
    > Perhaps it has crept into the pool by mistake.
    >
    > I don't see why ntpdate is quite so frowned upon.
    >


    You mean issues like this don't bother you? Using ntpd instead avoids
    many of this issues brought up in this discussion.

    Danny
    _______________________________________________
    questions mailing list
    questions@lists.ntp.isc.org
    https://lists.ntp.isc.org/mailman/listinfo/questions


  20. Re: uk pool problem

    Danny Mayer wrote:

    > Ronan Flood wrote:
    >
    >>Heiko Gerstung wrote:
    >>
    >>
    >>>I could think of some kind of misuse-protection, we discussed this
    >>>several times. If a client polling rate is too high, send it the wrong
    >>>time...
    >>>
    >>>Probably this is just an attempt of the owner of this server to reduce
    >>>the load caused by mad clients. I just issued a sequence of ntpdate
    >>>calls and it seems that I am blocked by that server automatically after
    >>>a couple of requests (for a short time).

    >>
    >>This appears to be the same server described at
    >>
    >> http://ntp.lp0.eu/
    >>
    >>ntp.lp0.eu is a CNAME to blackhole.arlott.org.uk.
    >>
    >>Quote:
    >>
    >>"Access
    >> o This server is private use only.
    >> o All other users should use pool.ntp.org.
    >> o Anyone using ntpdate (or equivalent) or unauthorised frequent
    >> polling of the time will be reported to their ISP's abuse contact.
    >> Any queries should be directed to the hostmaster (at blackhole.arlott.org)."
    >>
    >>Perhaps it has crept into the pool by mistake.
    >>
    >>I don't see why ntpdate is quite so frowned upon.
    >>

    >
    >
    > You mean issues like this don't bother you? Using ntpd instead avoids
    > many of this issues brought up in this discussion.
    >
    > Danny
    > _______________________________________________
    > questions mailing list
    > questions@lists.ntp.isc.org
    > https://lists.ntp.isc.org/mailman/listinfo/questions
    >


    Is there an ntpd equivalent to ntpdate -du ?

+ Reply to Thread
Page 1 of 4 1 2 3 ... LastLast