Python CM-API Example to pull Impala Query metrics

impalaQueries.py

#!/usr/bin/python

## *******************************************************************************************
##  impalaQueries.py
##
##  Getting Info on Impala Queries
##
##  Usage: ./impalaQueries.py  
##
## *******************************************************************************************


## ** imports *******************************

import sys
import pprint
from datetime import datetime, timedelta
from cm_api.api_client import ApiResource

fmt = '%Y-%m-%d %H:%M:%S %Z'

pp = pprint.PrettyPrinter(indent=4)

## ** Settings ******************************

## Cloudera Manager Host
cm_host = "toronto"
cm_port = "7180"

## Cloudera Manager login
cm_login = "admin"

## Cloudera Manager password
cm_password = "admin"

## Cluster Name
cluster_name = "Cluster 1"

## *****************************************

impala_service = None

## Connect to CM
print "\nConnecting to Cloudera Manager at " + cm_host + ":" + cm_port
api = ApiResource(server_host=cm_host, server_port=cm_port, username=cm_login, password=cm_password)

## Get the Cluster 
cluster = api.get_cluster(cluster_name)

## Get the IMPALA service
service_list = cluster.get_all_services()
for service in service_list:
  if service.type == "IMPALA":
    impala_service = service
    print "Located Impala Service: " + service.name
    break
    
if impala_service is None:
  print "Error: Could not locate Impala Service"
  quit(1)

## I'll configure this example to use a window of one day 
now = datetime.utcnow()
start = now - timedelta(days=1)

print "Looking for Impala queries executed by the user \"mark\""

filterStr = 'user = mark'

impala_query_response = impala_service.get_impala_queries(start_time=start, end_time=now, filter_str=filterStr, limit=1000)
queries = impala_query_response.queries

for i in range (0, len(queries)):
  query = queries[i]
  if (query.statement != "GET_SCHEMAS") and not (query.statement[:3] =="USE") and not (query.statement[:4] =="SHOW"):
  
    print '-------------------------------------'
   
    print "queryState : " + query.queryState
    print "queryId: " + query.queryId 
    print "user: " + query.user
    print "startTime: " + query.startTime.strftime(fmt)
    print "SQL: " + query.statement
    pp.pprint(query.attributes)
    
print '-------------------------------------'
print "done"

how can i get database and table in same script? using queryId.. is there any query attributes to show database and table name?

Your CM API code is really useful.
I am looking for a similar code to monitor hive queries and CM API services.py doesn't have anything related to the hive. do you done anything on the hive ?