Making Archive Files
Hopper supports the creation of and operation on a number of archive file formats, including TAR, HTAR, ZIP, and JAR. Creation of these files is treated like a file transfer job, so see the file transfer help page for details on monitoring progress of these jobs. Creating an archive file can be done in a number of ways. Most simply, just drag the files and directories you want to collect and drop them into a directory icon or open space in a directory window. In the resulting drop-menu, choose "Make TAR Archive Here..." (e.g.)
If you're viewing files in a directory window and wish to create a TAR file from some of those files, just select the items of interest and drag them into white space of the directory window. The drop menu you'll see will look something like:
Selecting the tar option will bring up the create tar dialog:
Note that the correct file suffix will be added automatically, based on compression options you choose (e.g., ".tar", ".tar.gz", ".tar.bz2"). You can override this by supplying a suffix when entering the archive name.
A very powerful option when creating TAR archives is to stream the tar file contents to a remote host over an FTP connection. The result is a single TAR archive on the remote host, but with the advantage of it being done with just a single transfer. This can be 10 or more times faster than copying individual files due the speed up of aggregation. When creating a remote TAR like this, you must be able to connect to the remote host via FTP. Here's how the tar creation dialog looks when doing this operation (note the FTP port option, which most of the time you can ignore):
A variation on the remote TAR archive is the remote Indexed TAR archive. The only difference with this option is that an additional file, called an index, is created which catalogs the contents of the archive. The pairing of the archive and index files can be operated upon directly by the HTAR program.
When running Hopper on LC production hosts users have the ability to create HTAR archives. The mechanism is very much like for TAR archives as explained above. HTAR is a program provided as part of HPSS for doing very high-speed transfers of data to and from HPSS. The fast transfer rates are the result of aggregation (combining many small files into one larger one), plus parallelism.
If you find yourself moving small files to storage (HPSS), especially in large quantities, you should definitely consider using HTAR for these transfers. HTAR can be 10 to 50 or more times faster than transferring the same files individually.
With the current version of HTAR one should be careful about the size of archive being created. Archives that span multiple tapes are "legal", but not recommended due to the increase in the time it takes to retrieve data from such archives, plus the increased risk of being spread across multiple tapes. Archives of roughly 300GB or less are recommended. Note that a near-future version of HTAR will automatically split huge archives into families of files, so that no one segment will span a tape boundary.
Hopper can also create JAR and ZIP archives using the same mechanism as described for TAR and HTAR. For these archives, the source files and the resulting archive must be on the same host on which you're running Hopper.