class MongoStorageTest(unittest.TestCase): def setUp(self): uri = os.environ.get("WMA_MONGODB", "mongodb://localhost:8230") self.dbname = "test_fwjr" try: self.mgr = MongoStorage(uri, dbname=self.dbname) self.mgr.remove() except: self.mgr = None print("WARNING: cannot connect to %s" % uri) data = { "int": 1, "float": 1.2, "list": [1, 2, 3], "dict": {"dname": "foo", "dval": 1}, "listdict": [{"lname": "foo"}], "str": "string", } self.bare_data = dict(data) data["wmaid"] = wmaHash(data) data["stype"] = "mongodb" self.data = data def tearDown(self): "Tear down content of temp dir" self.mgr.remove() self.mgr.dropdb(self.dbname) def test_write(self): "Test write functionality" if self.mgr: wmaids = self.mgr.write(self.data) self.assertEqual(len(wmaids), 1) data = self.mgr.read(wmaids[0]) record = data[0] for key in ["wmaid", "stype"]: if key in record: del record[key] self.assertEqual(record, self.bare_data) data = self.mgr.read(wmaids[0], ["dict"]) self.assertEqual(1, len(data)) self.assertEqual(data[0]["dict"], self.bare_data["dict"])
def cleanup(muri, tst, stype): "Cleanup data in MongoDB (muri) for given timestamp (tst)" time0 = time.time() mstg = MongoStorage(muri) # remove records whose type is hdfsio, i.e. already migrated to HDFS, # and whose time stamp is less than provided one query = {'stype': stype, 'wmats':{'$lt': dateformat(tst)}} rdocs = mstg.ndocs(query) tdocs = time.time()-time0 print(tstamp('mongo2avro'), 'found %s docs (in %s) to be removed' % (rdocs, elapsed_time(time0))) time0 = time.time() response = mstg.remove(query) print(tstamp('mongo2avro'), 'remove query %s in %s' % (query, elapsed_time(time0)))
def cleanup(muri, tst, stype, verbose): "Cleanup data in MongoDB (muri) for given timestamp (tst)" time0 = time.time() mstg = MongoStorage(muri) # remove records whose type is hdfsio, i.e. already migrated to HDFS, # and whose time stamp is less than provided one query = {'stype': stype, 'wmats': {'$lt': dateformat(tst)}} if verbose: print("Clean-up records in MongoDB: %s" % muri) print("MongoDB cleanup spec:", query) response = mstg.remove(query) if verbose: print("response: %s" % response) print("Elapsed time: %s" % elapsed_time(time0))
def cleanup(muri, tst, stype, verbose): "Cleanup data in MongoDB (muri) for given timestamp (tst)" time0 = time.time() mstg = MongoStorage(muri) # remove records whose type is hdfsio, i.e. already migrated to HDFS, # and whose time stamp is less than provided one query = {'stype': stype, 'wmats':{'$lt': dateformat(tst)}} if verbose: print("Clean-up records in MongoDB: %s" % muri) print("MongoDB cleanup spec:", query) response = mstg.remove(query) if verbose: print("response: %s" % response) print("Elapsed time: %s" % elapsed_time(time0))
def cleanup(muri, tst, stype): "Cleanup data in MongoDB (muri) for given timestamp (tst)" time0 = time.time() mstg = MongoStorage(muri) # remove records whose type is hdfsio, i.e. already migrated to HDFS, # and whose time stamp is less than provided one query = {'stype': stype, 'wmats': {'$lt': dateformat(tst)}} rdocs = mstg.ndocs(query) tdocs = time.time() - time0 print(tstamp('mongo2avro'), 'found %s docs (in %s) to be removed' % (rdocs, elapsed_time(time0))) time0 = time.time() response = mstg.remove(query) print(tstamp('mongo2avro'), 'remove query %s in %s' % (query, elapsed_time(time0)))