Re: 1 Machine, 2 NICs, 2 Instances of ntpd;Possible? - NTP

This is a discussion on Re: 1 Machine, 2 NICs, 2 Instances of ntpd;Possible? - NTP ; > One instance of ntpd is all that is necessary to perform both of these > tasks at the same time. I realize that what I am trying to do is very easily accomplished with one instance of ntpd. However, ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: Re: 1 Machine, 2 NICs, 2 Instances of ntpd;Possible?

  1. Re: 1 Machine, 2 NICs, 2 Instances of ntpd;Possible?


    > One instance of ntpd is all that is necessary to perform both of these


    > tasks at the same time.



    I realize that what I am trying to do is very easily accomplished with
    one
    instance of ntpd. However, I assure you that the manner in which I am
    trying to accomplish it is absolutely necessary for my particular
    application.

    I found a definitive answer in the code.

    ntp_io.c, line 1232

    /*
    * Check to see if we are going to use the interface
    * If we don't use it we mark it to drop any packet
    * received but we still must create the socket and
    * bind to it. This prevents other apps binding to it
    * and potentially causing problems with more than one
    * process fiddling with the clock
    */

    This explains why an instance was still receiving packets from an
    interface
    that was not specified with '-I'.


    Thank you very much for the help.

    Cheers

    Johnny

  2. Re: 1 Machine, 2 NICs, 2 Instances of ntpd; Possible?

    John.Johnson@gdc4s.com (Johnson, John-P63914) writes:


    >> One instance of ntpd is all that is necessary to perform both of these


    >> tasks at the same time.



    >I realize that what I am trying to do is very easily accomplished with
    >one
    >instance of ntpd. However, I assure you that the manner in which I am
    >trying to accomplish it is absolutely necessary for my particular
    >application.


    Uh, and we believe you why? It is very hard to imagine a situation where
    what you want is "absolutely necessary".



    >I found a definitive answer in the code.


    >ntp_io.c, line 1232


    >/*
    > * Check to see if we are going to use the interface
    > * If we don't use it we mark it to drop any packet
    > * received but we still must create the socket and
    > * bind to it. This prevents other apps binding to it
    > * and potentially causing problems with more than one
    > * process fiddling with the clock
    > */


    >This explains why an instance was still receiving packets from an
    >interface
    >that was not specified with '-I'.



    >Thank you very much for the help.


    >Cheers


    >Johnny


  3. Re: 1 Machine, 2 NICs, 2 Instances of ntpd; Possible?

    On 2008-03-11, Johnson, John-P63914 wrote:

    > I realize that what I am trying to do is very easily accomplished with
    > one instance of ntpd. However, I assure you that the manner in which I
    > am trying to accomplish it is absolutely necessary for my particular
    > application.


    Can you tell us why this is so?

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

  4. Re: 1 Machine, 2 NICs, 2 Instances of ntpd;Possible?

    Johnson, John-P63914 wrote:
    >> One instance of ntpd is all that is necessary to perform both of these

    >
    >> tasks at the same time.

    >
    >
    > I realize that what I am trying to do is very easily accomplished with
    > one
    > instance of ntpd. However, I assure you that the manner in which I am
    > trying to accomplish it is absolutely necessary for my particular
    > application.
    >


    There is something here that you are not telling us.

    > I found a definitive answer in the code.
    >
    > ntp_io.c, line 1232
    >
    > /*
    > * Check to see if we are going to use the interface
    > * If we don't use it we mark it to drop any packet
    > * received but we still must create the socket and
    > * bind to it. This prevents other apps binding to it
    > * and potentially causing problems with more than one
    > * process fiddling with the clock
    > */
    >
    > This explains why an instance was still receiving packets from an
    > interface
    > that was not specified with '-I'.
    >


    Not quite. It's more complicated than that. There are other addresses
    involved like the wildcard addresses and the loopback addresses.

    Danny

+ Reply to Thread