[evla-sw-discuss] NTP, EVLA MIB's and phase jumps

[James Robnett]

I've been continually befuddled by what I'm seeing.  I tried
removing what appeared to be the 'bad' reference server thinking
it was causing confusion.

Technically that shouldn't have helped since the two servers
would just ignore it in favor of the two Stratum-1's.  Oddly
enough it didn't help.

We're seeing large variances in time and I can't find anything
fundamentally different in how we have NTP configured so I
decided to check the kernel which is the other major thing that
changed in the OS upgrade.

Sure enough:
http://www.ntp.isc.org/bin/view/Support/KnownOsIssues#Section_9.2.4.2.7.

Redhat 5.3 is based off of the 2.6.18 kernel which this issue
comments upon.

Both Amos and Andy are using TSC to keep time.

I need to test the solution on another machine (probably that
odd ball server) and if it results in an improvement I'll need
to restart one of either Amos or Andy and then the other.

james