gemini-k8s-cluster

#!/bin/bash
#
# Convenient script to run gemini on cluster
#
# N.B:
# make sure kubectl has the same version as server

# variables
declare -r DB_HOST='scylladb'
declare -r SPARK_MASTER='spark://p-spark-master:7077'

# create pod if not exists
if ! kubectl get pods | grep gemini > /dev/null; then
  kubectl create -f ./k8s/gemini-pod.yaml || exit 1
fi

while ! kubectl get pods | grep "gemini.*Running" > /dev/null; do
  sleep 1
done

case $1 in
  hash)
    kubectl exec gemini -- /bin/bash -c "MASTER=$SPARK_MASTER /gemini/hash -h $DB_HOST ${@:2}"
    ;;
  query)
    kubectl exec gemini -- /gemini/query -h $DB_HOST "${@:2}"
    ;;
  report)
    kubectl exec gemini -- /gemini/report -h $DB_HOST "${@:2}"
    ;;
  bash)
    kubectl exec gemini -it /bin/bash
    ;;
  *)
    cat >&2 << END;
Available commands:
hash [options] <path> - hashes given set of Git repositories, either from container's FS or HDFS
query [options] <path> - finds duplicate file among hashed repositories
report [options] - shows all duplicated files among hashed repositories
bash - opens bash inside gemini container on cluster
END
    ;;
esac