Red Hat Bugzilla – Bug 79067
Ntpd does not work with non-fixed interfaces
Last modified: 2007-04-18 12:48:52 EDT
Description of Problem:
Ntpd seems to look only once, at startup time, to the local addresses for
binding purposes. This means that you run into trouble with for example
ADSL/cable/dial-up PPP interfaces and with VPN tunnels, of which the interface
and address is available after ntpd startup. NTP packets will then be generated
using a wrong source address (e.g., this might be a private address), which
might not work, either because of routing issues or by firewall blocks.
Version-Release number of selected component (if applicable):
4.1.1a-9 (applies to all versions I know).
Steps to Reproduce:
1. Set up a system with an 192.168.x.y address.
2. Do not enable masquerading.
3. Start ntpd.
4. Make a PPP connection with a default route.
5. Let ntpd talk to a public NTP server.
Packets will be sent out with a 192.168.x.y address.
Packats are sent out with the external PPP address.
how about /sbin/service ntp restart in /etc/ppp/ip-up.local ??
I will not fix this. Please mail this "Request for Extension" to email@example.com, thx.
Your ip-up.local hint is what already I did (as well as in
/etc/cipe/ip-up.local), although I used "condrestart" ;-). But a more
structural solution would be better. Anyway, I will post the bug to firstname.lastname@example.org.