wiki:JobStage

Version 2 (modified by davea, 12 years ago) (diff)

--

Staging input files

Before using an input file you must "stage" it, i.e. put it where it is accessible via HTTP from volunteer computers. The default way to do this is to store it on your BOINC server, in your project's download directory hierarchy. To do this, run

bin/stage_file [--gzip] [--copy] file_path
--gzip
send the file in compressed form to 7.0+ clients. Note: you must also include the <gzip/> attribute for this file in the job's input template.
--copy
copy the file from its current location to the BOINC download directory. The default is to move it.

stage_file also

  • checks whether a file of that name already exists in the download hierarchy. If so, it verifies that the file has the same contents.
  • computes and saves the file's MD5 hash. This eliminates the need to do this each time a job submission refers to the file.

Note: stage_file was added to the BOINC trunk on 16 Oct 2012. If your server code is older than that, use

cp file_path `bin/dir_hier_path file_name`

You may also stage input files to servers other than your BOINC server. In that case you must specify the URLs in the input template.