[time] 8 hour jump in time
ntpadmin at oc9.com
Mon Dec 4 13:54:09 UTC 2006
On Mon, 4 Dec 2006, Claypole, David wrote:
> THIS MESSAGE IS SUBJECT TO THE CONFIDENTIALITY NOTICE AT THE END OF THIS COMMUNICATION
> Have been using the uk.pool.ntp.org service for some time now without
> any issues and its been great. On saturday at 09:30 our server sync'd
> and jumped by 8 hours causing us a few issues. Just windering if anyone
> else had issues and if there is anything i can do to ensure it doesnt
> happen again.
> David Claypole
Are running ntpd ? If not I would recommend it.
Basically, if you aren't running ntpd, you must be using some kind of
sntp client ( e.g. ntpdate on linux, w32time on windows).
The sntp clients ( s is for simple) take their time from the first
available server as far as I can tell. This raise the risk of what happend
to you happening again. Any given server may go wrong (any clock
device also), not only servers from the pool.
When you run ntpd you usually specify 3 or more time source servers to
connect to :
ntpd is smart enough to detect if one of the 3 servers go wrong and ntpd
will continue to provide you with reliable time unless all servers go
If you do not run ntpd, consider setting at least 1 ntpd server for your
network and have your local machines read time from your own ntpd server.
It is OK to use sntp client on your internal network because you
can now trust your own ntpd server to give the right time.
If you are already running ntpd with multiple servers as source of time,
then something more serious did occur beacuse all the time source servers
would have to fail in order for an 8 hour glitch to occur.
More information about the pool