ntp usage on a machine out of sync
 

Just to check my sanity. We had a machine which was 7 mins out, due to ntp peers being blocked by firewall. After the issue was cleared, I brought, the time to within 2 mins, using DATE command. Then I used ntpdate command several times to get the time approximately right. Once it was right, I enabled ntp. I was told by my peers that I was wrong, and it would affect the Oracle database. The best way to do this is the let the machine come to the correct time on it's own. I don't understand that, having installed 100s of machines, am I missing something? What happens during summer/winter time changes, when an hour change occurs in one step? Does it not affect the database?

If it is 7 minutes AHEAD then setting it back MIGHT cause issues especially on a RAC cluster if both nodes don't have the same time. However if it is 7 minutes behind setting it forward isn't likely to cause issues.

Oracle stores a variable on its own that relates to time so I'm not sure it would even matter setting it AHEAD at OS. This particular variable is the source of a bug reported by Oracle a couple of years back though I forget the details.

By the way you can use the ntpdate -q <ntpserver> syntax to go ahead an update to current time rather than mucking with the date command.

p.s. Make sure you use hwclock to sync your hardware clock with your system clock so you'll start with correct time on next boot.