On 11 October 2016 at 17:02, alfonso alfonso.ros-dos-santos@evosoft.com wrote:

Hello everyone,

just a simple question. I might be wrong but I understand that submitting a job through lava-dispatch and lava-tool should lead to the same process.

Yes, if your templates are correct. This is one of the main ways to debug problems with the templates.

Now, with the dispacher you can already specify an specific target device encoded in yaml format.

https://staging.validation.linaro.org/static/docs/v2/pipeline-debug.html#com...

Does the lava-tool at some point reach a similar target configuration? does it generate it or it is stored somewhere? in the latter case, where is it stored?

It is generated from the templates on the master and sent to the worker. What the worker sees is the same in each case. You can also re-generate the output using lava-tool with the get-pipeline-device-config command.

The device YAML is written to /tmp/lava-dispatcher/slave/ in a directory based on the job ID. (This is subject to change in future.)

lava-tool uses XML-RPC to submit to the master.