Closed btalb closed 2 years ago
This appears to be an issue in documentation. The following from the main benchbot page does not work
benchbot_submit --containerised --example hello_active
however this neglects updated functionality for containerised mode and so should be updated to the following
benchbot_submit --example-containerised hello_active
However, when testing this I did find that the --containerised
option does not seem to work for examples unless being run from within the folder of the Dockerfile.
Example:
u@pc: benchbot/addons/benchbot_addons/benchbot-addons/examples_base/examples/.hello_benchbot$ benchbot_submit -c ./Dockerfile
works fine but
u@pc: benchbot/addons/benchbot_addons/benchbot-addons/examples_base/examples$ benchbot_submit -c .hello_benchbot/Dockerfile
does not
Thanks for your testing @david2611 . I've run command:
u@pc: benchbot/addons/benchbot_addons/benchbot-addons/examples_base/examples/.hello_benchbot$ benchbot_submit -c ./Dockerfile
exactly under the folder as above, and shell is able to execute the docker build
command properly, building a temporate image. But then there shows that "non-network local connections being added to access control list, waiting for Docker network to become available" with no further information, which seems is a Docker-related problem. How can I fix it?
@TongChiYip is that behaviour happening even when you have a benchbot_run
command going?
If not it will hang there, looking for a Docker network until one is created via the benchbot_run
command.
@david2611 Thanks for your reply. Command
u@pc:~$ benchbot_run --robot carter_omni --env miniroom:1 --task semantic_slam:active:ground_truth
runs well on my system. Does it mean that I have to run a benchbot_run
command at the same time as I run the benchbot_submit
command?
Standard pipeline for operating with benchbot is as follows:
benchbot_run
command in a given terminal. This is where you define what task you are performing, what robot you are using, and the environment you are operating withinbenchbot_submit
in a separate terminal while benchbot_run
is still going. This enables your solution to interact with the environment as you attempt to complete the experiment you are running. benchbot_eval
command to see how well you did after your solution has completed. This will not require benchbot_run
to be running and must happen after your solution has completed (benchbot_submit
has ended).More info can be found on the main benchbot page and in our Tutorial.
In your situation what I think you are wanting to do is the following:
u@pc:~$ benchbot_run --robot carter_omni --env miniroom:1 --task semantic_slam:active:ground_truth
u@pc: benchbot/addons/benchbot_addons/benchbot-addons/examples_base/examples/.hello_benchbot$ benchbot_submit -c ./Dockerfile
Now the hello_benchbot
example is an example that is meant to have the robot drive backwards and forwards again to test everything is working ... but in miniroom it will crash in hilarious ways :) so if you want to have an example that shows you benchbot more thoroughly I might recommend running the hello_active
example instead
Appears as though examples don't work in
--containerised
mode.Will investigate and post a fix.
For now,
--native
mode should work (or running without the flag altogether).