Re: [ldm-users] dvb1.ssec clock problems??`

On 10/15/10 15:45, dc wrote:
Hi

First thing we checked was that ntp was running and it was. Next checked
to make sure ntp.conf matched dvb2, and it did. Restarted ntpd service
and about 4 minutes later dvb1 was synced up.

We are not sure why ntp stopped working.


You say it was running but you say it stopped working. How did you diagnose the two issues?

My experience with NTP over the last decade or so is that it's a very reliable daemon. Basically, once it's set, it's set and your clock will always be withing a few ms of UTC.

You need to have multiple timesources. Three really is the minimum to allow NTP to run it's algorithms properly to estimate delays due to the network. More is better, though I'm sure there is a point of diminishing returns.

If you run the command below, you can see how many remote systems your local system is using for time. If the "reach" column is not always or almost always "377" (an octal value) for a given source, then you probably should look at another source of time (if it's off your campus) or at your network connectivity (if it's on your campus).

[root@ldmingest02 ~]# ntpq -pn
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
*10.197.0.6      10.197.1.253     2 u  209  512  377    0.171   -1.517   0.901
+10.197.0.24     10.197.1.253     2 u  253  512  377    1.458   -0.381   0.178
+10.197.0.23     10.197.1.253     2 u  260  512  377    0.183   -1.335   0.142

DON'T use the local clock as a source!!!! If you use RHEL or it's spawn, delete these lines:


# Undisciplined Local Clock. This is a fake driver intended for backup
# and when no outside source of synchronized time is available.
server  127.127.1.0     # local clock
fudge   127.127.1.0 stratum 10


If, for some reason, you reboot your server frequently, you can speed convergence of NTP by adding "iburst" to the server lines of your config. You should also have a "driftfile" line.

In fact, most of the lines in the default RHEL configuration can go. Mine look like this (yes, the "server" line will resolve a name, but the "restrict" line won't - go figure):


[root@ldmingest02 ~]# cat /etc/ntp.conf
# VIA ks.cfg ...
restrict default ignore
restrict 127.0.0.1
#
server ntp01 iburst prefer
restrict 10.197.0.6 mask 255.255.255.255
#
server ntp02 iburst
restrict 10.197.0.24 mask 255.255.255.255
#
server ntp03 iburst
restrict 10.197.0.23 mask 255.255.255.255
#
driftfile /var/lib/ntp/drift


HTH.


--
Peter Laws / N5UWY
National Weather Center / Network Operations Center
University of Oklahoma Information Technology
plaws@xxxxxx
-----------------------------------------------------------------------
Feedback? Contact my director, Craig Cochell, craigc@xxxxxx. Thank you!



  • 2010 messages navigation, sorted by:
    1. Thread
    2. Subject
    3. Author
    4. Date
    5. ↑ Table Of Contents
  • Search the ldm-users archives: