[evlatests] news

[Bill Sahr]

Please see below.

Ken Sowinski wrote:
> On Tue, 5 Jun 2007, Ken Sowinski wrote:
> 
>> During one of Rick's observations early this afternoon and for
>> a while afterward it appears that the EVLA antennas' idea of
>> time seemed to differ from the rest of the VLA for while.  I
>> planned to disable phase switching to pin down the source of
>> the problem, but while discussing it with Jim Jackson it
>> magically cured itself.  This raises the obvious question of
>> whether the changed time multicasting contributed to this.
>> Without further evidence I doubt it.  The same thing occurred
>> for a while last week, albeit after many serious disturbances
>> to the system.  Should the antennas continue to lose sync with
>> the VLA over the next few days I think we should consider turning
>> off the multicast of time from andy.
> 
> James explained to me that coincidentally there was a machine in
> Socorro multicasting at the time Monday afternoon when the L302
> mibs appeared to be confused about time.  There is no doubt that
> these packets would have made it to the EVLA antennas.

I take it that it is safe to assume that this machine in Socorro
was multicasting time, and further in a valid NTP format ?  I ask
because I have been experimenting with multicasting using mnemosyne
and orca, but only text messages, not time, and to a port I
believe to be unused by the EVLA (multicast addr 239.192.0.1,
port 20050).  I cannot remember if I was running this software on
Monday.

> 
> While not a clear explanation it this seems a likely cause.  There
> is still no reason to discontinue having both amos and andy multicast
> time, but we should remain alert to possible timing errors.
> 
> James and I discussed the merits of letting the VLA network use the
> GPS at the VLA as its primary NTP server.  This would insulate us
> from drifting time if the network to the outside world is lost.
> I believe it is possible to configure NTP so that it still looks at 
> the outside time servers so that a failure of the GPS will be detected
> and reported.  Is this a good idea?

The notion of using the GPS at the VLA site as the primary NTP server
seems sound, as does the idea of checking the GPS time against external
time servers and reporting significant differences between the two
time sources.

Bill Sahr
> 
> _______________________________________________
> evlatests mailing list
> evlatests at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evlatests