[ntp:questions] systems won't synchronize no matter what

David Lord snews at lordynet.org
Sat Oct 30 16:50:32 UTC 2010


David J Taylor wrote:
> "David Lord" <snews at lordynet.org> wrote in message 
> news:t06tp7-n0b.ln1 at p4x2400c.home.lordynet.org...
> []
>> You've hit on why it's difficult to have a generic ntp.conf.
>> It's not just ntpd version specific but operating system version
>> specific.
>>
>> I swapped my NetBSD systems to FreeBSD after NetBSD-1.6x as I
>> couldn't get NetBSD-2 to work on my hardware. I swapped back
>> to NetBSD-3.x when I had hardware compatibility problems with
>> FreeBSD.
>>
>>
>> The default NetBSD ntp.conf for NetBSD 3.1 4.x and 5.1_RC4
>> are same:
>>
>> pidfile    /var/run/ntpd.pid
>> driftfile  /var/db/ntp.drift
>> logconfig  -syncstatus
>> tos        minsane 2
>> server     0.pool.ntp.org
>> server     1.pool.ntp.org
>> server     2.pool.ntp.org
>>
>>
>> My own ntp.conf has restrict lines to allow config from
>> certain IP addresses and statistics logging enabled.
>>
>> Local peers have 'minpoll 6 maxpoll 8 iburst' and are back
>> in sync within about five minutes from a restart. If all
>> peers are down, as a few weeks back when there was a BT
>> MSO, it was more like an hour before ntp was ticking again
>> (I need another couple of radioclocks to cover for this).
>>
>> Remote servers have 'minpoll 9 maxpoll 11' except from
>> ISPs I use which have shorter minpoll and 'iburst'.
>>
>>
>> David
> 
> David,
> 
> This is growing out of all proportion to my original remark!  I simply 
> commented on the lack of an example ntp.conf file in a document. 
> Something like:
> 
>  server  0.pool.ntp.org  iburst
>  server  1.pool.ntp.org  iburst
>  server  2.pool.ntp.org  iburst
> 
> would have done nicely as a sample, on which you can build.  Like you, I 
> have changed minpoll and maxpoll to suit my own needs, and added various 
> other commands.  Perhaps you would agree that's it's easier to have 
> something working to start with, which you can then edit later as you 
> understand more of how things work in practice.
> 
> Anyway - any clue on:  Why add  maxpoll 9 to all the servers?  That 
> seems wrong.  I see the note: "The option `maxpoll 9' is used to prevent 
> PLL/FLL flipping on FreeBSD.".  So is that a FreeBSD specific issue, and 
> if so, why not fix it in the OS?

Was that not fixed in a later release?

(I don't follow FreeBSD but there were similar issues at
different times with NetBSD and Linux with kernel
timekeeping that were fixed in subsequent releases)

David




More information about the questions mailing list