Bug 56122: Original bug report misunderstood; root problem not corrected
Hint #1 is the subject of the original report: Race condition between ntp and NetworkManager at entry to runlevel 2 Hint #2 from original report: Scope: The symptom I see is a problem with ntp. The underlying cause could affect other network services as well. Hint #3 from original report: log excerpt showing ntp trying to initialize 5 seconds before NetworkManager brings up eth0. Briefly: As long as NetworkManager brings down the network interface and then brings it up again, all asynchronously to the startup of network services, some of those network services are sometimes going to try to initialize themselves while the network interface is down and consequently fail. It is impractical to try to bulletproof every network service to deal with the network interface being down when the service tries to start. No network service should be started until NetworkManager indicates that the network interface is up. I have tried to explain this as clearly as I can. I don't have the energy to rehash it again. If whoever reads this doesn't understand what a race condition is and why Debian network startup has a race condition please kick this message upstairs. Thanks.
