YOHKOH archive tapes contain a tape directory, reformatted data and software. Tapes are created on DIGITAL's Unix (Ultrix) system using `cp', `cpio' and `tar' archiving utilities. Reformatted data files and the weekly files are all copied to tape using the `cpio' utility. The software archive is copied to tape with the `tar' utility. Since, the software archive comprises several different directory trees the `tar' utility was the logical choice because it allows one preserve these directory structures during restore operations. Archive tapes will nominally contain a single calendar week (always beginning with Sunday) of reformatted data. (Where each reformatted file contains data from a single orbit which spans approximately 90 minutes.) This calendar week is given a number (week number: 1-53) which is then encoded into the name of the tape in the format `yy_wka.xx', where `yy' is the year, `wk' is the week number, `a' is the media number (always ``a'' for tapes, but can be ``b'' for other media, such as MO disks), `xx' is the revision number or level of the archive tape.
The detailed contents of a YOHKOH data archive tape begins with two directory files in different formats (ASCII and binary), followed by a series of `cpio' data archive files, and lastly a `tar' software archive file.
Structure of the data and software archive tape. Each cpio archive file contains only those reformatted data that share the same file ID (e.g. archive file 2 contains files from the file ID: 911201.1545, file 3 would contain files with file ID 911201.1715).
The tape directory file lists the contents of each tape. The file contains general information such as when the tape was made, the name of the tape, the number of `cpio' archive files, the first and last data file identification (fileID) contained within the `cpio' archive, and a breakdown on the contents of the software `tar' archive file giving names and sizes of embedded `tar' files. The tape directory also has two detailed tables. The first of which tabulates the weekly files present within the first `cpio' archive and the second tabulates the `orbit' fileID and the sizes of each reformatted file (e.g. ada, bda, cba, hda, sfr, spr, wda) for each `cpio' archive file. This last table is very useful for identifying whether a given file is on a tape.
The reformatted data are placed on tape with the help of the Unix archiving utility `cpio' which allows a specified list of files to inserted into a single archive file. Using this utility we have bundled the data by placing files with the same data file identification or fileID (such as 911201.1654, which corresponds to the date 1-Dec-91 at 16 hours and 54 minutes) together into a single archive file and copied to tape. Each fileID is unique and represents one orbit of data. Since there are about 16 orbits per day, each tape could have 112 + 1 different `cpio' archive files. The additional archive file is for the first `cpio' file which only contains the weekly files such as the observing log file (obs*) and the pointing file (pnt*).
Even though the software is placed on the tape, it is best to obtain the software across the network. See Appendix A for more details.