def getQueryParams(args):
print("Obtaining query params from the database.")
client = tsquery.createQueryClient(args.endpoint, profile=args.profile)
## Use the following query to get the query parameters from the data
## by picking attributes each type of time series to initialize the dimensions.
## Introduces some randomization to enable multiple executions to pick different parameters.
queryStr = """
WITH selectedTaskCompleted AS (
SELECT *
FROM "{0}"."{1}"
WHERE measure_name = 'task_completed'
AND time BETWEEN {2} - 2h AND {2}
LIMIT 100
), selectedCpu AS (
SELECT t.*
FROM "{0}"."{1}" t INNER JOIN selectedTaskCompleted c ON c.instance_name = t.instance_name
WHERE t.measure_name = 'cpu_user'
AND t.time BETWEEN {2} - 2h AND {2}
ORDER BY random()
LIMIT 1
)
SELECT * FROM(
(
SELECT * FROM selectedCpu
)
UNION
(
SELECT * FROM selectedTaskCompleted
)
)
ORDER BY measure_name
""".format(args.databaseName, args.tableName, args.queryEndTime)
print(queryStr)
result = tsquery.executeQueryAndReturnAsDataframe(client, queryStr, True)
params = Params(args.databaseName, args.tableName, result['region'][0],
result['availability_zone'][0], result['cell'][0],
result['silo'][0], result['microservice_name'][0],
result['instance_type'][0], result['os_version'][0],
result['instance_name'][0], result['process_name'][1],
result['jdk_version'][1])
return params
def __init__(self,
threadId,
args,
startTime,
queryProvider,
tableName=None):
threading.Thread.__init__(self)
self.threadId = threadId
self.args = args
self.startTime = startTime
self.config = configparser.ConfigParser()
self.config.read(args.config)
retries = int(
self.config.get(configDefaultSection, configRetries, fallback=0))
self.client = tsquery.createQueryClient(region=args.region,
profile=args.profile,
retries=retries,
endpoint=args.endpoint)
self.output = ()
self.tps = 0.0
self.queryCount = 0
self.tableName = tableName
## Query parameters
if self.args.fixedParams:
## Use fixed query parameters
self.params = Params(
self.args.databaseName, self.args.tableName, 'eu-west-1',
'eu-west-1-1', 'eu-west-1-cell-8', 'eu-west-1-cell-8-silo-1',
'apollo', 'r5.4xlarge', 'AL2',
'i-ojtoEEXU-apollo-eu-west-1-cell-8-silo-1-00000033.amazonaws.com',
'host_manager', 'JDK_8')
else:
## Read the query parameters from the database and table.
self.params = getQueryParams(self.args)
print(self.params)
queryInstances = queryProvider(self.params, args.queryEndTime,
args.wide)
## Initialize the query mode.
self.queryMode = self.config.get(configDefaultSection,
configQueryMode,
fallback=configQueryModeRegular)
if self.queryMode not in [
configQueryModeRegular, configQueryModeRowCount
]:
raise Exception('Unknown query mode: {}'.format(self.queryMode))
if self.queryMode == configQueryModeRegular:
self.queries = getQueryStrings(queryInstances, self.params)
else:
queriesRowCount = createQueryInstancesRowCount(
queryInstances, args.queryEndTime)
self.queries = getQueryStrings(queriesRowCount, self.params)
## Initialize the query distributions.
if configQueryDistributionSection not in self.config:
raise Exception('Query distribution section missing')
queryDistributions = self.config[configQueryDistributionSection]
distributions = dict()
for query in queryDistributions:
if query not in self.queries:
raise Exception('Unknown query: {}'.format(query))
if query in distributions:
distributions[query] += float(queryDistributions[query]) / 100
else:
distributions[query] = float(queryDistributions[query]) / 100
self.queriesSelected = list()
self.queryWeights = list()
sum = 0
for query in distributions.keys():
print('{} : {}'.format(query, distributions[query]))
self.queriesSelected.append(query)
self.queryWeights.append(distributions[query])
sum += distributions[query]
print("Sum of probabilities: ", sum)