How Spica auto works ?

The default installation path for spica is /usr/local/spica. So all scripts are located in this directory.
Spica should be installed on each node of a cluster so you need to install the package as many times as the number of hosts in your cluster. Running Spica doesn't affect other node so to launch a whole set of reduction on a cluster you'll need to launch spica on each node.

On each node, when you run the script spica_auto.pl a set of shell scripts is build. An email is sent when the job is complete.
Each file will run a specific processing. They have all .cmd extension. CFHTLS_W_i_022929-060403_T0001.cmd is for processing a CFHTLS field (Wide, filter i, RA : 02h29m29s Dec : -06h04'03'')

A launching script is also built for each stack (.spica extension). A deamon (spica_daemon.pl) will run them one by one. When a stack is completed, an email is sent and the launching script move from the spica extension to the completed extention. If an error occur, the launching script have the error extension.

The web page : http://clix.iap.fr/spica/admin/admin-spica/survey/ show :

1. processes running
2. list of pending processes.

Running Spica

Once the QualityFITS have been done on all images and they have been ranked, you are ready to launch the stacking. Spica will search in the database which files the stacks require for calibration and stacking. Only images ranked 'A' or 'B' will be used.
The first step is to build your own configuration file (see Spica configuration for help). Then run spica_auto.pl -c . If the spica daemon is running, processing will start immediately, if not, just launch the script spica_daemon.pl

To see the list of stack produced, have a look at : http://clix.iap.fr/spica/admin/admin-spica/survey/list_stack.php The set of images for the astrometric calibration and stacking are shown.

You can view the current processing step at : http://clix.iap.fr/spica/admin/admin-spica/survey/ for stack http://clix.iap.fr/spica/admin/admin-spica/ for image by image

Spica extension file

When the list of stack have been build, the files have the extension .spica. Once started, the extension is moved to .current

spica extension	status
.spica	pending
.current	processing
.processed	completed
.error	error occur

Dealing with errors

There are many ways to stop the processing : 1) machines power failure 2) software crash 3) stopping the spica deamon

In case one, the script can't stop properly and so can't save its step. To restart, you will need to delete the current stacks and relaunch spica_auto.pl to rebuilt them.

For case two, once you have resolve the software failure, you just need to change the status of the stack by moving its extension from .error to .spica. The process will then restart from the failure command. Of course, if the problem is still there, Spica will move back the file extention from .current to .error.

Case three doesn't require anything from you because the current processing stack is beeing completed.

Deleting a stack

Select the stacks you want to delete with : http://clix.iap.fr/spica/admin/admin-spica/survey/

Hints

 File descriptor should be changed to a higher values (done for our cluster) or you won't be able to process more than a few dozens of MegaCam files.