1126870 - [tracker] Improving NTP for releng

Status: RESOLVED FIXED

(Infrastructure & Operations :: Infrastructure: Other, task)

Description

[Jake Watkins [:dividehex]]

As it currently stands, each of our data centers have at least 2 ntp services which are peered with each other.  We really should be peering all of our ntp servers across the board.  It should also not include itself as a peer.

In addition to those changes, the ntp config should also allow query from localhost.  It just makes it easier to inspect without typing the fqdn while logged into the host.

Comment 1

[Jake Watkins [:dividehex]]

Also forgot to mention ntp within the releng BU.  Those should be also peered with the rest of our internal ntp infra.

Comment 2

[Jake Watkins [:dividehex]]

Another note here; since we are running ntpd on vm guests, we shouldn't have a local clock fallback.

Comment 3

[Brian Hourigan [:digi]]

Just wanted to check in - how's NTP looking? Is there anything left to do here?

Comment 4

[Jake Watkins [:dividehex]]

(In reply to Brian Hourigan [:digi] from comment #3)
> Just wanted to check in - how's NTP looking? Is there anything left to do
> here?

I think we are good here. I don't think we have seen any issues with ntp for awhile now. Thanks, Digi

Comment 5

[Jake Watkins [:dividehex]]

And if we see any issues again in the future, I'm going to just blame it on ntpd running under vms and immediately point all of releng to hardware ntpd in scl3.  :-P