mstrechen/parcs-example

This is an example of problem solution based on language agnostic PARCS implementation (see lionell/parcs)

How to run

First, you need to run PARCS on your cluster. For this step, please use lionell's manual.

Next, just use your leader machine (with $ gcloud compute ssh leader) and run something like:

$ LEADER=tcp://<leader address in cluster>:4321
$ IMAGE=mstrechen/largest-common-subseqence-runner-py:latest
$ WORKERS=<number of workers>
$ sudo docker service create \
        --name runner \
        --network parcs \
        --restart-condition none \
        --env LEADER_URL=$LEADER \
        --env WORKERS=$WORKERS \
        --env N=100000 \
        --env M=300 \
        --env SEED=10 \
        $IMAGE

Use swarmpit ui to get worker/leader logs :)

Running your own image

For this you have to use your own DockerHub repository. Something like:

docker build -t <your_repo>/largest-common-subseqence-runner:latest .
docker push <your_repo>/largest-common-subseqence-runner:latest

Now you are able to run your own runner with the same command as before - just set your own IMAGE variable.

For worker replacement you also have to build & push an image, but at the same time don't forget to change image name in the code of runner (see task = self.engine.run('mstrechen/largest-common-subseqence-worker-go') line in main.py)