It turns out that those awg scripts don't release their exitations nicely, which means that we weren't going to be able to go to Observe until they were cleared.  In clearing them, I accidentally stopped the CW hardware injections. Once those were back in place, we're back to Observing.

The rest of this alog is some details, in case we need to refer to them long-term.

When trying to figure out how to stop an individual excitation (the Transient hardware excitation point) I misunderstood the meaning of the channel numbers on the GDS TABLE from the calinj's GDS screen (see attachment), and inadvertently stopped the CW injection rather than the Transient injection path.  Those numbers are for *testpoints*, not awg slots (which is consistent with what the top row of the table says, but it didn't click for me).  Erik reminded me later that I could have checked awg slots with diag> awg show 42 (where 42 is h1calinj's number), but it's still not clear that you can selectively clear a single excitation.  Jamie and the CDS folks are going to follow this up.

In the end, I did an awg clear 42 * and also tp clear 42 *, and that released all the excitations (since we already had to restart the CW injections).

Apparently there is a 'monit' process that should automatically restart the CW injections if they are stopped for some reason like this.  However, Dave found that since tconvert has an output (due to a warning), that monit process was not being successful. Dave worked some magic, and tconvert no longer gave a warning, which meant that the monit process was able to restart the CW injections. After that was done, we were able to go back to Observe long term.

Until we understood the issue with the startup script, I had been holding us out of Observe since we didn't have the CW injections. Keith let me know that they have monitoring downstream for when those are or are not in place, so it would have been fine if the CW injections were not in place for a few hours to a ~day, until we were able to debug.  I tried to set us to Observing, however every few minutes the auto-startup-process was trying to start the injection, which begins with setting a gain to zero (so that it can later be ramped on), so that SDF diff kept popping us out of Observe.  Dave stopped the monit process, so we went back to Observe and stayed there for a few minutes.  By then, Dave had fixed the tconvert warning issue, and we went back out of Observe one last time, Dave restarted the monit process, and it restarted the CW injections.  Now we're *really* back in Observe.

EDIT: Dave's alog about the magic: alog 70775