| 1 |
[Sorry if this not entirely coherent: it's growing late here.] |
| 2 |
|
| 3 |
On Wednesday 03 April 2013 22:45:11 Mick wrote: |
| 4 |
|
| 5 |
> Do you have chronyc set up to signal off/online status to chronyd? If |
| 6 |
> so, where do you run it from? |
| 7 |
|
| 8 |
No, I don't need to bother with manual control: it just works for me. I did |
| 9 |
play with chronyc years ago but I soon got bored. |
| 10 |
|
| 11 |
I have chronyd running on my LAN server, getting its time from several hosts |
| 12 |
Out There, then internal hosts get their time from the LAN server. All very |
| 13 |
straightforward. Any time errors just ramp themselves steadily down to zero. |
| 14 |
I can show you my config files if you like. |
| 15 |
|
| 16 |
In those old 486/586 days the author used to publish a new version for each |
| 17 |
new generation of CPU, but I haven't seen that since x86_64 came on the |
| 18 |
scene. I suppose he must have found another way of calibrating his object |
| 19 |
code's execution time. Whatever the case, I'm happy that gkrellm etc. show |
| 20 |
the time to at least the nearest half-second relative to BBC time signals. |
| 21 |
|
| 22 |
I started using chrony in those old dial-up days because it can handle slow |
| 23 |
links. Also intermittent ones, by which I mean systems that shut down for |
| 24 |
some of the day, or that dual-boot with Windross,. I haven't felt a need to |
| 25 |
change since then. At that time ntpd was at a disadvantage IIRC. You might |
| 26 |
say that chrony is designed for desktop boxes rather than the servers that |
| 27 |
ntpd likes, though the edges are pretty blurred. |
| 28 |
|
| 29 |
-- |
| 30 |
Yours, happy camper of Tideswell, |
| 31 |
Peter |
Archives