keep changing the way I write back up data disks so the number written in August is not comparable to the July total, But it is much larger. At least 200 by the old method. The important number now is the number of star measurements. This was 14,870,000 through the 30th with a small chance for part of one more night. But no luck. It is raining so I gave up on breaking 15Million and am sending in my monthly report.
Everything is running smoothly. I have started cleaning up the production operation. I have written a script to backup the collected images. The old way was to manually click on each file to be saved using xcdroast. Since a good evening produces over a thousand files, this was a lot of clicking, and was subject to error. It was not easy to select the files that should be saved, so I always saved too many. Now I have written a script which looks at the .clt files and saves the .fits files that produced each .clt file. This saves just the files that are in the final result. There is still a bug in the script but it writes correct data files.
I have a dual processor 1.5 GHz machine processing TOM1 and TOM2, and a 1 GHz processing TOM3. I usually get up at dawn, close the dome, and start the processing. The scheme is that each pipeline processor copies the files from the camera machines into a buffer area and starts the computations. This way I have 24 hours to complete the computations. A good night will run nearly the full 24 hours and if there are any complications like the network being flaky, the process can really back up. I have ordered another dual processor machine, and will probably split things up so that 6 machines process the three cameras.
I have written another script that takes a big cal file and the WS file produced from it. I can make a cut on the WS file like WS > 5 or some such. The program looks at the WS file, and looks up the stars in the big .cal file. It then deletes a selected number of outliers and checks that the resulting data is some number of sigma above what is expected at the star's magnitude. It then writes out a .png plot file, a data file and a star list in VizieR format for each of the stars that met the cut.
This produces a list where the majority of the stars are real variables.
Dan has fabricated a shipping crate, and we will shortly send out a set of upgrades which I hope will get ARNE running again.