Re: Re: [chrony-dev] Issue about chronyd synchronize to local clock

[ Thread Index | Date Index | More chrony.tuxfamily.org/chrony-dev Archives ]


   I opened the built-in log in chronyd and after several hours the synchronization status becomes no. But  the stratum didn't increase too high, as I use

   local stratum 10 in chrony.conf as its initial value, the straum just increase by 1.

    chronyd[10693]: ntp_core.c:1113:(transmit_timeout) Transmit timeout for [127.0.0.1:123]
   chronyd[10693]: ntp_io.c:828:(NIO_SendPacket) Sent 48 bytes to 127.0.0.1:123 from [UNSPEC] fd 4
   chronyd[10693]: ntp_io.c:666:(process_message) Received 48 bytes from 127.0.0.1:51122 to 127.0.0.1 fd=3 if=1 tss=1 delay=0.000035102
   chronyd[10693]: clientlog.c:480:(CLG_LogNTPAccess) NTP hits 67 rate -40 trate -128 tokens 0
   chronyd[10693]: ntp_io.c:828:(NIO_SendPacket) Sent 48 bytes to 127.0.0.1:51122 from 127.0.0.1 fd 3
   chronyd[10693]: ntp_io.c:666:(process_message) Received 48 bytes from 127.0.0.1:123 to 127.0.0.1 fd=4 if=1 tss=1 delay=0.000021056
   chronyd[10693]: ntp_core.c:1686:(receive_packet) NTP packet lvm=44 stratum=11 poll=10 prec=-24 root_delay=0.000031 root_disp=0.034988   refid=7f000001 []
   chronyd[10693]: ntp_core.c:1691:(receive_packet) reference=1593144591.569475574 origin=545624252.384833077 receive=1593177798.926692774 transmit=1593177798.926771957
  chronyd[10693]: ntp_core.c:1693:(receive_packet) offset=0.000002204 delay=0.000018974 dispersion=0.000000 root_delay=0.000049 root_dispersion=0.034989
  chronyd[10693]: ntp_core.c:1696:(receive_packet) remote_interval=0.000079183 local_interval=0.000098157 response_time=0.000079183 txs=D rxs=K
  chronyd[10693]: ntp_core.c:1700:(receive_packet) test123=111 test567=111 testABCD=1110 kod_rate=0 interleaved=0 presend=0 valid=1 good=0 updated=1



 
From: Bill Unruh
Date: 2020-06-26 01:06
To: chrony-dev
Subject: Re: Re: [chrony-dev] Issue about chronyd synchronize to local clock
Miroslav explained. Because a client's stratum is one higher than the server
it gets time from. Since it is getting time from itself, each time it does so,
its stratum increases by 1. When it gets to 16 it stops using a server with
that high a stratum. Other servers do not keep increasing their stratum.
 
 
On Thu, 25 Jun 2020, xqmeng80@xxxxxxxxxxx wrote:
 
>    So why chronyd can always sucessfully synchronize to other server but it can't always synchronize to itself
>
> ________________________________________________________________________________________________________________________
>        
> From: xqmeng80@xxxxxxxxxxx
> Date: 2020-06-25 23:19
> To: chrony-dev
> Subject: Re: Re: [chrony-dev] Issue about chronyd synchronize to local clock
>    I have a server cluster made of a number of machines, and I just choose one from them to be the NTP server.
> then all other machines synchronize time to
>
>   the server. My application runs on every machine of the cluster, and the app will not work if it detected 'NTP
> not synchronized status', so I have to make
>  
>    the machine which act the role the NTP server synchronize to itself, but the status of the machine will be not
> synchronized after several hours and will not become synchronized any more.
>
> ________________________________________________________________________________________________________________________
>        
> From: Bill Unruh
> Date: 2020-06-25 23:10
> To: chrony-dev
> Subject: Re: Re: [chrony-dev] Issue about chronyd synchronize to local clock
>  
> On Thu, 25 Jun 2020, xqmeng80@xxxxxxxxxxx wrote:
>  
> >    Is there a way or some configuration to make chronyd client always successfully synchronize to itself.
>  
> Why? Your database program is designed to only work if the system clock is
> synchronized to UTC. It seems that you want to break that, and wnat to use
> chrony to help you do that. The directive rtcsync should switch on that flag
> when the system clock really is synchronized. Is that not what you want? Or
> you can take Miroslav's advice and periodically (eg every minute if necessary)
> use adjtimex to switch off that flag. Or hack the kernel code to never
> switch it off. Or hack the database code to stop checking on that flag.
>  
> >
> >
> _____________________________________________________________________________________________________________________
> >        
> > From: Miroslav Lichvar
> > Date: 2020-06-25 22:54
> > To: chrony-dev
> > Subject: Re: Re: [chrony-dev] Issue about chronyd synchronize to local clock
> > On Thu, Jun 25, 2020 at 10:44:29PM +0800, xqmeng80@xxxxxxxxxxx wrote:
> > >    As the application runs on the NTP server machine, time on the machine is thinked accurate. I just
> can't
> > figure out why synchonizing to itself will make
> > >    the NTP synchronization status becomes No after several hours.
> >  
> > Because the NTP stratum will increase with each iteration in the loop
> > and the client will stop synchronizing when it reaches 16. Then
> > it will just drift and when the root dispersion is too large, the
> > clock will be considered "unsynchronized".
> >  
> > --
> > Miroslav Lichvar
> >  
> >  
> > --
> > To unsubscribe email chrony-dev-request@xxxxxxxxxxxxxxxxxxxx with "unsubscribe" in the subject.
> > For help email chrony-dev-request@xxxxxxxxxxxxxxxxxxxx with "help" in the subject.
> > Trouble?  Email listmaster@xxxxxxxxxxxxxxxxxxxx.
> >  
> >
> >
> >
>
>
>


Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/