[evlatests] Software time February 12

[Barry Clark]

Software time started late because the SLC was in the process of dying
(or at least being very sick), and it seemed best to get that diagnosed
rather than starting software time.

When I tried to bring up idcaf, the system sat there and sulked for about
twenty minutes, then suddenly started working all on its own.

I did a preliminary job of setting delays.  Delays are set to within about
10 ns for IFs A, C, and D.  The delay solver used antenna 24 as reference,
and something seemed to be wrong with IF B on that antenna, and no 
satisfactory solutions were available.  These delays were set
using modes 1A, 1C, 1D.  Fringes didn't look very satisfactory when I
tried mode 4, but I didn't get around to investigating why.

Midway through the above, I had to stop idcaf.  When I restarted it, it
once more sulked for about twenty minutes, and then started all on its own.

I tried running idcaf on igloo, to send data to the archive, but I discovered
that the relevant multicast from the executor was not routed to igloo.
Idcaf said it was getting data from the correlator but didn't know what
to do with it, so the twenty minute pause did not occur here, nor did it
occur when I restarted idcaf on mchost.

When we switched back to Modcomp control, the correlator controler had to
be rebooted; not clear why.

Idcaf occasionally inserted the message that it hadn't received data from
the correlator, and had therefore suppressed a dump.  I didn't investigate
whether this was in fact the case, or whether the message was being produced
in error.


To summarize the process of invoking the Modcomp free system, the steps 
for doing so are very straightforward, but they don't work very well.