(use a fixed font!)
I have a 3 node QNX 4 system, with node 3 acting as a stratum 4 server and
the other 2 nodes taking the time via NTP from it. The version of NTP is
3.5f, which is the latest according to the QNX support site. All seems to be
fine, with the 3 clocks synchronising, yet if left overnight (or around 5 hours), the time on
the 2 client nodes drifts and NTP appears to stop :
root / >on -n1 date ; on -n2 date ; on -n3 date
Fri Jul 07 08:13:35 utc 2000
Fri Jul 07 08:13:32 utc 2000
Fri Jul 07 08:13:36 utc 2000
This is despite the fact that DMPEERS in XNTPDC shows this for node 3
xntpdc> dmpeer
remote local st poll reach delay offset disp
*LOCAL(0) 127.0.0.1 4 64 377 0.00000 0.000000 0.01781
ntpserver 5.0.0.0 16 64 0 0.00000 0.000000 16.0000
and this for for both nodes 1&2
xntpdc> dmpeer
remote local st poll reach delay offset disp
LOCAL(0) 127.0.0.1 10 64 377 0.00000 0.000000 0.01781
*ntplocal2 192.168.201.101 5 64 377 0.00000 0.055386 0.01050
+ntplocal1 192.168.200.101 5 64 377 0.00000 0.054926 0.01059
ntpserver 5.0.0.0 16 1024 0 0.00000 0.000000 16.0000
(ntplocal1&2 are 2 routes to node 3)
According to this, my ntp server is synched to the local clock, and the
client is synched to the ntp server.
When looking at the log for the client, the last entry occurs at 19.30ish :
6 Jul 19:11:23 xntpd[18074]: synchronized to LOCAL(0), stratum=10
6 Jul 19:11:33 xntpd[18074]: synchronized to 192.168.200.103, stratum=5
6 Jul 19:26:30 xntpd[18074]: time reset (step) 1.001123 s
6 Jul 19:26:30 xntpd[18074]: synchronisation lost
6 Jul 19:31:09 xntpd[18074]: synchronized to 192.168.201.103, stratum=5
First it syncs to the local clock, then almost straight away it syncs with
the NTP server (it does this every time). Then once its synced the clock,
sync is lost with the NTP server (it does this EVERY time it changes the
clock). That last sync to the NTP server is the last entry in the log.
My ntp.conf for node 3 (the server) looks like this :
New NTP Configuration file for node 3
6/7/2000 Initial Revision
server ntpserver prefer
server 127.127.1.0 #If all else fails, use local clock
fudge 127.127.1.0 stratum 4 #Set to 4, which is not high, but higher than
local clock on nodes 1&2
(The server ‘ntpserver’ does not respond at the moment - this would be the
customer configuration)
My ntp.conf for node 1 (not serving anyone else) looks like this :
New NTP Configuration file for node 1 & 2
6/7/2000 Initial Revision
server ntpserver prefer
server ntplocal1
server ntplocal2
server 127.127.1.0
fudge 127.127.1.0 stratum 10
(The server ‘ntpserver’ does not respond at the moment - this would be the
customer configuration)
The idea behind this configuration is that the customer will have an NTP
server of their own, which all nodes will sync to, but as a fall back nodes
1&2 will sync to node 3, and if all else fails they`ll use their own clocks.
This configuration seems to work fine for about 6/7 hours, then it stops!
Has anyone had similiar problems using NTP (3.5f) and QNX4?
Thanks in advance!
Richard…
\
Richard Levy - Software Engineer
Telspec Europe Ltd
Richard.Levy@Telspec.Co.Uk