cloudera-managerizer

#!/usr/bin/env python

import getopt, sys, os, csv

def usage():
    print "%s [-v] [-g] --region <region>" % sys.argv[0], \
    "--key-name <key-name> --image-id <image-id> --db-pass <db-pass>", \
    "[--cluster <cluster.csv>]"

def main():
    try:
        opts, args = getopt.getopt(sys.argv[1:], "hr:k:i:c:d:vg",
        ["help", "region=", "key-name=", "image-id=", "cluster=", "db-pass="])
    except getopt.GetoptError as err:
        # print help information and exit:
        print str(err)  # will print something like "option -a not recognized"
        usage()
        sys.exit(1)

    verbose = False
    generate_only = False
    region = None
    key_name = None
    db_pass = None
    image_id = "ami-c74789a9" # default image: centos 7.2
    cluster = "cluster.csv" # default csv name

    for o, a in opts:
        if o == "-v":
            verbose = True
        elif o == "-g":
            generate_only = True
        elif o in ("-h", "--help"):
            usage()
            sys.exit()
        elif o in ("-r", "--region"):
            region = a
        elif o in ("-k", "--key-name"):
            key_name = a
        elif o in ("-i", "--image-id"):
            image_id = a
        elif o in ("-c", "--cluster"):
            cluster = a
        elif o in ("-d", "--db-pass"):
            db_pass = a
        else:
            assert False, "unhandled option"

    if region == None or key_name == None:
        usage()
        sys.exit()

    if verbose:
        print "region:   ", region
        print "key_name: ", key_name
        print "image_id: ", image_id
        print "cluster:  ", cluster
        print "db_pass:  ", db_pass

    rows = []
    with open (cluster) as csvfile:
        reader = csv.reader(csvfile, delimiter=',')
        for row in reader:
            if len(row) <= 8 or (len(row[0].strip()) > 0 and row[0].strip()[0] == '#'):
                # skip illegal entry or comment
                continue
            modrow = [r.strip() for r in row]
            modrow[-1] = " ".join([ "Key=%s,Value=%s" % tuple(pair.split('=')) for pair in modrow[-1].split(' ') ])
            rows.append(modrow)
    hosts = ["%s   %s" % (row[1], row[2]) for row in rows]

    generator_base = "./generator.sh -r %s -h %s -c '%s'"
    launch_instance_aws = "./launch-instance-aws.sh %s %s %s %s"
    for row in rows:
        generator = generator_base
        if row[0] == "server":
            generator += " -p " + db_pass
        # build user-data shell
        os.system(generator % (row[0], row[2], "\n".join(hosts)))
        if not generate_only:
            os.system(launch_instance_aws % (region, image_id, key_name, ' '.join(row)))
            os.system("rm cloudera-managerizer-*.sh")

if __name__ == "__main__":
    main()