Description of Problem: My linux gateway machine serves both named and ntp, I noticed the other day that after a reboot of the server as ntp started it failed to synch with my upstream servers (initscripts running ntpdate to make the clock close). I discovered that ntp is started by default prior to named, and since the system uses it's own name server for resolving addresses, ntp failed to find the upstream servers causing the ntp service to die. My recommended fix is to reorder the initscripts to start named almost immediately after the network comes up, prior to any other network services that may rely on correct resolving of DNS addresses.
Perhaps ntp should start later?
That's reasonable of course, but I was thinking of any other services that might desire/require name resolution at startup (now or in the future).
[harald@faro src]$ fgrep chkconfig /etc/init.d/ntpd # chkconfig: - 58 74 [harald@faro src]$ fgrep chkconfig /etc/init.d/named # chkconfig: - 55 45
Argh. I reported this bug against the wrong version. It's only screwed up on Red Hat Linux 7.2/Alpha, not in Limbo. My bad. Gotta quit staying up this late. Sorry. # grep -e chkconfig /etc/init.d/named # chkconfig: - 55 45 # grep -e chkconfig /etc/init.d/ntpd # chkconfig: - 26 74
Closing bugs on older, no longer supported, releases. Apologies for any lack of response.