Add restart_rssi action to pysmurf controller

[tristpinsm]

Using caput to directly access the register here is not ideal and it would probably be better to wrap this in a function in pysmurf. What do you think?

Pending simonsobs/smurf_dockers#4 and simonsobs/smurf-streamer#43

[tristpinsm]

We should probably also tag and upgrade sodetlib as long as we're at it so that fixed tones and other features are included.

[tristpinsm]

I wasn't able to test the new action on the SLAC system without disrupting the running OCS agents.

[msilvafe]

Looks good to me.

[BrianJKoopman]

Agreed that a wrapped function in pysmurf for this would be better than the S._caput() call.

Any plans to test this somewhere, even at site, before we merge?

One comment below, but not super critical to add.

[tristpinsm]

Note that this is an opportunity to deploy some new features in sodetlib. Merging simonsobs/sodetlib#457 and bumping the tag before releasing this version is something we should consider.

[tristpinsm]

Agreed that a wrapped function in pysmurf for this would be better than the S._caput() call.

Any plans to test this somewhere, even at site, before we merge?

One comment below, but not super critical to add.

re testing: I haven't worked with OCS before and I was trying to figure out how to test it. I think you would have to update the code checked out by the crossbar and restart everything (?) This seemed like it could be disruptive

[BrianJKoopman]

re testing: I haven't worked with OCS before and I was trying to figure out how to test it. I think you would have to update the code checked out by the crossbar and restart everything (?) This seemed like it could be disruptive

It'll sort of depend on your lab setup. You could bring down a pysmurf-controller agent there, and start up this modified agent in its place, then run the new task and verify functionality. This certainly is disruptive if you're actively using the pysmurf-controller. Rolling back once testing is complete is generally straight forward though.

For starting up the modified agent you can either build the image with the modified code in it, or run it directly on the host via something like:

$ cd $OCS_CONFIG_DIR
$ OCS_CONFIG_DIR=$(pwd) ocs-agent-cli --site-file=default.yaml --instance-id=pysmurf-controller-id --site-host=hostname

You'll need to change pysmurf-controller-id and hostname to suite your setup.

Building the docker image, you'd need to update the docker-compose.yml file with the new image tag and run a docker compose up -d (or down/up if you're using Host Manager.) This is the more thorough option, as it would catch any issues with how the software within the image runs, useful especially if the underlying image is going to change with those potential sodetlib changes.

[tristpinsm]

OK, thanks for the suggestions. Starting the pysmurf-controller docker with the updated code is not an issue. I just don't know how to trigger the new command. The way things are set up there is a "crossbar" server running on another host -- which does not have updated software. So if I spawn a client there it doesn't know about the new restart_rssi command.

[BrianJKoopman]

Starting the pysmurf-controller docker with the updated code is not an issue. I just don't know how to trigger the new command.

From any host on the network that can access the crossbar server you can run:

$ ocs-client-cli shell <agent instance-id>

That will drop you into a python session with an OCSClient object called client. You can then run client.restart_rssi(). (This assumes the default.yaml is properly configured and OCS_CONFIG_DIR is set.)

The way things are set up there is a "crossbar" server running on another host -- which does not have updated software. So if I spawn a client there it doesn't know about the new restart_rssi command.

No changes are needed on the crossbar server. The crossbar server is a communication layer between the client you're spawning with the above command and the agent. So only the agent code needs to be updated as in my last post.

[tristpinsm]

Thanks! That was very helpful and in the end I was able to test the new agent task and make some fixes.

I was able to run the command using ocs-client-cli and then confirm that the register had been updated in rogue.