AWS Kinesis shard split

kinesis_split.py

import time
from boto import kinesis

AWS_PROFILE = None
AWS_REGION = 'us-east-1'
AWS_STREAM_NAME = 'YourStreamName'

WAIT_STATUS_SEC = 5
WAIT_STATUS_ROUNDS = 5

options_split = True
options_shard_num = 14

k = kinesis.connect_to_region(AWS_REGION, profile_name = AWS_PROFILE)

stream_info = k.describe_stream(AWS_STREAM_NAME)

shards = {}
shards_num = len(stream_info['StreamDescription']['Shards'])

for shard in stream_info['StreamDescription']['Shards']:
  s_id = shard['ShardId']
  shards[s_id] = {
    'StartingHashKey': int(shard['HashKeyRange']['StartingHashKey']),
    'EndingHashKey':   int(shard['HashKeyRange']['EndingHashKey']),
  }

  if options_split:
    status = None
    i = 0
    while status != 'ACTIVE':
      status = k.describe_stream(AWS_STREAM_NAME)['StreamDescription']['StreamStatus']
      if i > WAIT_STATUS_ROUND or status == 'ACTIVE':
        break
      i += 1
      time.sleep(WAIT_STATUS_SEC)

    new_hkey = int((shards[s_id]['StartingHashKey'] + shards[s_id]['EndingHashKey']) / 2)
    print "Splitting shardId ", s_id, "- new hash: ", new_hkey

    k.split_shard(AWS_STREAM_NAME, s_id, str(new_hkey))
    shards_num += 1

    if options_shard_num == shards_num:
      break

stream_info = k.describe_stream(AWS_STREAM_NAME)
print "Shards after split: ", len(stream_info['StreamDescription']['Shards'])