The most vital characteristic of the CALDB structure is the indexing system. The index works with the header specifications in the various data files to ensure unambiguous file selection.
Special FITS files named caldb.indx contain the lists of CALDB keywords for the principle header data unit (HDU) of each calibration file; there is one index file for each instrumental branch. The caldb.indx files are actually links to the latest versioned files in the appropriate index directory; see the Versioning page for more information.
The index file format is prescribed in the HEASARC reference "Calibration Index Files" (CAL/GEN/92-008).
The columns included in each index file are listed in Table 1. There is one line in the table for each data set (each HDU) that has been entered into the index. Some files have only one HDU indexed; others, such as the ACIS bad pixel files, have one extension for each detector chip.
Format and length
|TELESCOP||10A||TELESCOP header keyword|
|INSTRUME||10A||INSTRUME header keyword|
|DETNAM||20A||DETNAM header keyword|
|FILTER||10A||FILTER header keyword|
|CAL_DEV||20A||Logical link necessary for CALDB access|
|CAL_DIR||70A||Subdirectory location of file|
|CAL_FILE||80A||CALDB filename being indexed|
|CAL_CLASS||3A||CCLS0001 keyword value|
|CAL_DTYP||4A||CDTP0001 keyword value|
|CAL_CNAM||20A||CCNM0001 keyword value|
|CAL_CBD||630A70||CBDn0001 keyword values|
|CAL_XNO||I||Extension number within the fits file of the dataset|
|CAL_VSD||10A||CVSD0001 keyword value (date only)|
|CAL_VST||8A||CVST0001 keyword value|
|REF_TIME||8-byte double (D)||CVSD0001 converted into modified Julian date|
Calibration quality value, 0=good, 5=bad;
see the Quality Keyword section.
|CAL_DATE||10A||Date the HDU was indexed in the CALDB|
|CAL_DESC||70A||CDES0001 keyword value|
Only two values of the quality keyword, CAL_QUAL, are currently used in the Chandra CALDB: 0 and 5. Good quality is indicated by the value 0. A value of 5 indicates that the calibration data has been replaced by better dataset.
Only the good files can be retrieved by a CALDB query. There must be only one good file for a given full specification, otherwise multiple datasets would be returned. Files marked as bad are kept as options for custom processing requests or for manual selection by the user.
Adding new files to the database is the responsibility of the Chandra CALDB manager. These files will be announced in an official patch to the CALDB and will be posted on the Download page. When you install the update, it will automatically write the calibration files, create a new versioned index file, and update the caldb.indx link for each affected branch.
You should not attempt to index your own files. If the local index files are changed, updates from the download page may not occur properly. Additionally, your internal changes will be overwritten by the CALDB upgrade.
It is also the responsibility of the CALDB manager to inform users when files have been retired from the CALDB. In general, it is unnecessary to remove files from the database unless there are disk space issues; files are simply marked as bad instead (see the Quality Keyword section). Retired files will be removed from the CALDB tarfiles on integral version updates (e.g. 3.0, 4.0).