20100805 when to fix the time in an oracle as infrastructure - plembo/onemoretech GitHub Wiki

title: When to fix the time in an Oracle AS infrastructure link: https://onemoretech.wordpress.com/2010/08/05/when-to-fix-the-time-in-an-oracle-as-infrastructure/ author: lembobro description: post_id: 133 created: 2010/08/05 10:56:02 created_gmt: 2010/08/05 10:56:02 comment_status: open post_name: when-to-fix-the-time-in-an-oracle-as-infrastructure status: publish post_type: post

When to fix the time in an Oracle AS infrastructure

If time ever gets out of sync on an Oracle AS infrastructure cluster, that’s a severe enough situation that you should feel comfortable just going ahead and making the fix one server at a time.

Where the delta is significant, consider restarting the apps.

For anyone who hasn’t done this keep in mind that the ntp service should first be shut down before running ntpdate for the update.

/etc/init.d/ntpd stop

That’s the Red Hat way.

Then try doing (as root) ntpdate [time server IP or name] (e.g. ntpdate time.example.com).

If your time is really way off you may need to use the date command to reset the clock.

Once you’re done, restart ntpd.

/etc/init.d/ntpd start.

Verify it’s up with /etc/init.d/ntpd status.

Then make sure it’s actually working by doing

/usr/sbin/ntpq

ntpq >pe

If you get anything back other than a good result for the time server(s) listed in your ntp.conf, it’s probably not configured correctly.

If this is an intranet system you should always use an internal time source that’s synchronized with another Stratum 2 server where available.

Even for Internet connected systems I’m not a big fan of using the one of the “free” pools. I’ve seen too many instances where a pool server was off just enough to cause headaches. if you’re going to direct connect to something on the Internet use a pair of reliable servers like those here in the States sponsored by or associated with NIST (e.g. utcnist.colorado.edu and/or time-a.nist.gov and time-b.nist.gov).