def fetch_splits_from_shards(config, uri):
"""Internal method to fetch splits from shareded db
:returns: The splits
"""
logging.warn(
"WARNING getting splits that connect directly to the backend mongods is risky and might not produce correct results"
)
connection = get_connection(uri)
configDB = connection["config"]
shardsColl = configDB["shards"]
shardSet = []
splits = []
cur = shardsColl.find()
for row in cur:
host = row.get('host')
slashIndex = host.find("/")
if slashIndex > 0:
host = host[slashIndex + 1:]
shardSet.append(host)
splits = []
for host in shardSet:
new_uri = get_new_URI(uri, host)
config['input_uri'] = new_uri
splits += calculate_unsharded_splits(config, new_uri)
#I think this is better than commented way
return splits
'''
def fetch_splits_from_shards(config, uri):
"""Internal method to fetch splits from shareded db
:returns: The splits
"""
logging.warn("WARNING getting splits that connect directly to the backend mongods is risky and might not produce correct results")
connection = get_connection(uri)
configDB = connection["config"]
shardsColl = configDB["shards"]
shardSet = []
splits = []
cur = shardsColl.find()
for row in cur:
host = row.get('host')
slashIndex = host.find("/")
if slashIndex > 0:
host = host[slashIndex + 1:]
shardSet.append(host)
splits = []
for host in shardSet:
new_uri = get_new_URI(uri,host)
config['input_uri'] = new_uri
splits += calculate_unsharded_splits(config,new_uri)
#I think this is better than commented way
return splits
'''
def __init__(self, stream, params):
config = {}
for key, value in params.__dict__.iteritems():
config[key] = value
self.uri = config.get("output_uri")
self.conn = get_connection(self.uri)
self.coll = get_collection(self.uri)
self.key_name = config.get("job_output_key", "_id")
self.value_name = config.get("job_output_value")
def __init__(self, stream, params):
config = {}
for key, value in params.__dict__.iteritems():
config[key] = value
self.uri = config.get('output_uri')
self.conn = get_connection(self.uri)
self.coll = get_collection(self.uri)
self.key_name = config.get('job_output_key', '_id')
self.value_name = config.get('job_output_value')
def __init__(self,stream,params):
config = {}
for key, value in params.__dict__.iteritems():
config[key] = value
self.uri = config.get('output_uri')
self.conn = get_connection(self.uri)
self.coll = get_collection(self.uri)
self.key_name = config.get('job_output_key','_id')
self.value_name = config.get('job_output_value')
self.add_action = config.get('add_action', 'insert')
self.add_upsert = config.get('add_upsert', False)
def __init__(self, params):
config = {}
for key, value in params.get('mongodb', {}).items():
config[key] = value
self.uri = config.get('output_uri')
self.conn = get_connection(self.uri)
self.coll = get_collection(self.uri)
self.key_name = config.get('job_output_key','_id')
self.value_name = config.get('job_output_value')
self.add_action = config.get('add_action', 'insert')
self.add_upsert = config.get('add_upsert', False)
self.base_doc = config.get('base_doc', {})
def fetch_splits_via_chunks(config, uri, use_shards):
"""Retrieves split objects based on chunks in mongo
:returns: The splits
"""
originalQuery = config.get("query")
if use_shards:
logging.warn("WARNING getting splits that connect directly to the \
backend mongods is risky and might not produce correct results")
logging.debug("fetch_splits_via_chunks: originalQuery: %s" % originalQuery)
connection = get_connection(uri)
configDB = connection["config"]
shardMap = {}
if use_shards:
shardsColl = configDB["shards"]
cur = shardsColl.find()
for row in cur:
host = row.get('host')
slashIndex = host.find("/")
if slashIndex > 0:
host = host[slashIndex + 1:]
shardMap[row.get('_id')] = host
logging.debug("MongoInputFormat.getSplitsUsingChunks(): shard map is: %s" % shardMap)
chunksCollection = configDB["chunks"]
logging.info(configDB.collection_names())
query = bson.son.SON()
uri_info = uri_parser.parse_uri(uri)
query["ns"] = uri_info['database'] + '.' + uri_info['collection']
cur = chunksCollection.find(query)
logging.info("query is ", query)
logging.info(cur.count())
logging.info(chunksCollection.find().count())
numChunks = 0
splits = []
for row in cur:
numChunks += 1
minObj = row.get('min')
shardKeyQuery = bson.son.SON()
min = bson.son.SON()
max = bson.son.SON()
for key in minObj:
tMin = minObj[key]
tMax = (row.get('max'))[key]
#@to-do do type comparison first?
min[key] = tMin
max[key] = tMax
if originalQuery == None:
originalQuery = bson.son.SON()
shardKeyQuery["$query"] = originalQuery
shardKeyQuery["$min"] = min
shardKeyQuery["$max"] = max
inputURI = config.get("input_uri")
if use_shards:
shardName = row.get('shard')
host = shardMap[shardName]
inputURI = get_new_URI(inputURI, host)
splits.append(MongoInputSplit(
inputURI,
config.get("input_key"),
shardKeyQuery,
config.get("fields"),
config.get("sort"),
config.get("limit", 0),
config.get("skip", 0),
config.get("timeout", True),
config.get("slave_ok",False)))
# return splits in uri format for disco
return [s.format_uri_with_query() for s in splits]
def fetch_splits_via_chunks(config, uri, use_shards):
"""Retrieves split objects based on chunks in mongo
:returns: The splits
"""
originalQuery = config.get("query")
if use_shards:
logging.warn("WARNING getting splits that connect directly to the \
backend mongods is risky and might not produce correct results"
)
logging.debug("fetch_splits_via_chunks: originalQuery: %s" % originalQuery)
connection = get_connection(uri)
configDB = connection["config"]
shardMap = {}
if use_shards:
shardsColl = configDB["shards"]
cur = shardsColl.find()
for row in cur:
host = row.get('host')
slashIndex = host.find("/")
if slashIndex > 0:
host = host[slashIndex + 1:]
shardMap[row.get('_id')] = host
logging.debug("MongoInputFormat.getSplitsUsingChunks(): shard map is: %s" %
shardMap)
chunksCollection = configDB["chunks"]
logging.info(configDB.collection_names())
query = bson.son.SON()
uri_info = uri_parser.parse_uri(uri)
query["ns"] = uri_info['database'] + '.' + uri_info['collection']
cur = chunksCollection.find(query)
logging.info("query is ", query)
logging.info(cur.count())
logging.info(chunksCollection.find().count())
numChunks = 0
splits = []
for row in cur:
numChunks += 1
minObj = row.get('min')
shardKeyQuery = bson.son.SON()
min = bson.son.SON()
max = bson.son.SON()
for key in minObj:
tMin = minObj[key]
tMax = (row.get('max'))[key]
#@to-do do type comparison first?
min[key] = tMin
max[key] = tMax
if originalQuery == None:
originalQuery = bson.son.SON()
shardKeyQuery["$query"] = originalQuery
shardKeyQuery["$min"] = min
shardKeyQuery["$max"] = max
inputURI = config.get("input_uri")
if use_shards:
shardName = row.get('shard')
host = shardMap[shardName]
inputURI = get_new_URI(inputURI, host)
splits.append(
MongoInputSplit(inputURI, config.get("input_key"), shardKeyQuery,
config.get("fields"), config.get("sort"),
config.get("limit", 0), config.get("skip", 0),
config.get("timeout", True),
config.get("slave_ok", False)))
# return splits in uri format for disco
return [s.format_uri_with_query() for s in splits]
