[ntp:questions] question regarding NTP configuration for clusters, and "cluster time" stability

Unruh unruh-spam at physics.ubc.ca
Wed Sep 30 15:57:35 UTC 2009


"rotordyn at yahoo.com" <rotordyn1 at gmail.com> writes:


>> Well, have 4 of them be the internal top level, and have the rest of the
>> nodes use all 4 as their servers. Then make sure that if one looses
>> sync with UTC, it drops down in stratum so that the rest of the machines
>> will ignore it.

>That seems logical. I'll look into that, thank you.

>> >I can do that too, but I run into chicken and egg issues, in that the
>> >cluster software needs the nodes to be in synch before starting,
>> >so NTP must already be running.
>>
>> Uh, yes, the computer must run in =A0order for the computer clock to run.
>> That is Always true.

>That's not quite what I was saying. I have clustering software that
>has modest time synchronization requirements, in that all nodes'
>clocks in the cluster need to be within a second of each other.
>In the current implementation, the software will not start unless
>ntpd is running. I could add functionality to that software to
>modify ntpd configurations dynamically, based on cluster state.
>But since ntpd must be running prior to the cluster software
>starting, at a minimum I need to have a boot-strap configuration
>that allows ntpd to start up and function at least as well as it
>does in the current implementation, even if the config might be
>changed in response to a change in cluster state.

Start up ntpd with the -g option. Select 4 of your machines as the
"standard" (servers), and start them up first syncing them with -g to
some one standard time standard. Then after a min start up the others
with these as servers, and with -g so the clock steps if it is out by
more than 1/8 sec. 



More information about the questions mailing list