Processes commands

The process subcommands allow you to get details about an Arkindex process, and to re-run its workers on failed elements. However, these commands have been made mostly obsolete by workers activity, which allows you to track the progress of your process, see the elements that failed, and retry the process skipping elements that have already been successfully processed, from the Arkindex frontend.

Process Reports

The report subcommand outputs error statistics about a process, retrieved from it ml_report.json artifacts.

arkindex process report $PROCESS_ID

The output looks like this, with errors sorted by class:

11061 elements: 10575 successful, 486 with errors
    Errors by class
┃ Class       ┃ Count ┃
│ HTTPError   │   470 │
│ KeyError    │    15 │
│ ReadTimeout │     1 │

Optional arguments

The --run argument allows you specify which run of the process you wish to retrieve reports for; by default, the command retrieves information from the latest run.

arkindex process report $PROCESS_ID --run 2

You can change the commands' output using the following arguments:

  • --json: this outputs an object containing all the elements from all the ML reports that have at least one error, which you can then save to a file.
arkindex process report $PROCESS_ID --json > $PATH/TO/file.json
  • --verbose: outputs the full error messages and tracebacks with syntax highlighting.
arkindex process report $PROCESS_ID --verbose

Process recovery

The recover subcommand allows you to start a new process from the failed elements (with at least one error) of your process.

arkindex process recover $PROCESS_ID

This retrieves the process' ML reports, lists the failed elements, adds them to your Arkindex selection and creates a new configured process, to which you are given a link which takes you to the Arkindex frontend where you can configure and start your new process.

arkindex process recover $PROCESS_ID
[11:58:44] INFO     Loading first page on try 1/5
2 elements selected.
You may now configure and start a new workflow using the Arkindex frontend:$PROCESS_ID/configure

If you already had elements in you selection, the tool will ask your permission to replace them.

You can specify the worker run to target using the --run argument:

arkindex process recover $PROCESS_ID --run 1