rally doesn't return results from sh/bash script in benchmark

[andrew_shi]

I run benchmark VMTasks.boot_runcommand_delete (rally/plugins/openstack/scenarios/vm/vmtasks.py).
It creates instance and run in it some shell script. The script returns information in json-format. But I can't find script's results in rally report and database. There is only "vm.run_command_over_ssh" (execution duration time shell script) in rally report.
But when I modify shell script and it returns non json or error, rally return error in the report.

In the rally/plugins/openstack/scenarios/vm/vmtasks.py written, that it returns dictionary with keys `data' and `errors': data: dict, JSON output from the script, errors: str, raw data from the script's stderr stream

Where can I find script's results or how the script should return results?

[Alexander Maretskiy (maretskiy)]

https://review.openstack.org/#/c/320460/