def setUp(self):
super(TestFull_WaitQuery, self).setUp()
time_hacker.set_real_time()
self.testid = str(os.getpid() + random.randint(1, 1000))
self.modconf = Module({'module_name': 'LiveStatus',
'module_type': 'livestatus',
'port': str(random.randint(50000, 65534)),
'pnp_path': 'tmp/pnp4nagios_test' + self.testid,
'host': '127.0.0.1',
'name': 'test',
'modules': ''
})
self.init_livestatus(self.modconf)
def setUp(self):
super(TestFull_WaitQuery, self).setUp()
time_hacker.set_real_time()
self.testid = str(os.getpid() + random.randint(1, 1000))
self.modconf = Module({'module_name': 'LiveStatus',
'module_type': 'livestatus',
'port': str(random.randint(50000, 65534)),
'pnp_path': 'tmp/pnp4nagios_test' + self.testid,
'host': '127.0.0.1',
'name': 'test',
'modules': ''
})
self.init_livestatus(self.modconf)
def setUpClass(cls):
# temp path for mongod files :
# as you can see it's relative path, that'll be relative to where the test is launched,
# which should be in the Shinken test directory.
mongo_path = cls._mongo_tmp_path = tempfile.mkdtemp(dir="./tmp/", prefix="mongo")
mongo_db = os.path.join(mongo_path, "db")
mongo_log = os.path.join(mongo_path, "log.txt")
os.system("/bin/rm -rf %r" % mongo_path)
os.makedirs(mongo_db)
print("Starting embedded mongo daemon..")
sock = socket.socket()
sock.bind(("127.0.0.1", 0))
port = sock.getsockname()[1]
sock.close()
cls.mongo_db_uri = "mongodb://127.0.0.1:%s" % port
mongo_args = [
"/usr/bin/mongod",
"--dbpath",
mongo_db,
"--port",
str(port),
"--logpath",
mongo_log,
"--smallfiles",
]
mp = cls._mongo_proc = subprocess.Popen(
mongo_args, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, shell=False
)
print("Giving it some secs to correctly start..")
time_hacker.set_real_time()
# mongo takes some time to startup as it creates freshly new database files
# so we need a relatively big timeout:
timeout = time.time() + cls.mongod_start_timeout
while time.time() < timeout:
time.sleep(1)
mp.poll()
if mp.returncode is not None:
cls._read_mongolog_and_raise(mongo_log, mp, "Launched mongod but it's directly died")
sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
errno = sock.connect_ex(("127.0.0.1", port))
if not errno:
sock.close()
break
else:
mp.kill()
cls._read_mongolog_and_raise(
mongo_log, mp, "could not connect to port %s : mongod failed to correctly start?" % port
)
time_hacker.set_my_time()
def tearDownClass(cls):
mp = cls._mongo_proc
mp.terminate()
print("Waiting mongod server to exit ..")
time_hacker.set_real_time()
for _ in range(10):
time.sleep(2)
if mp.poll() is not None:
break
else:
print("didn't exited after 10 secs ! killing it..")
mp.kill()
mp.wait()
os.system("/bin/rm -rf %r" % cls._mongo_tmp_path)
def tearDownClass(cls):
mp = cls._mongo_proc
mp.terminate()
print('Waiting mongod server to exit ..')
time_hacker.set_real_time()
for _ in range(10):
time.sleep(2)
if mp.poll() is not None:
break
else:
print("didn't exited after 10 secs ! killing it..")
mp.kill()
mp.wait()
os.system('/bin/rm -rf %r' % cls._mongo_tmp_path)
def setUpClass(cls):
# temp path for mongod files :
# as you can see it's relative path, that'll be relative to where the test is launched,
# which should be in the Shinken test directory.
mongo_path = cls._mongo_tmp_path = tempfile.mkdtemp(dir="./tmp/", prefix="mongo")
mongo_db = os.path.join(mongo_path, 'db')
mongo_log = os.path.join(mongo_path, 'log.txt')
os.system('/bin/rm -rf %r' % mongo_path)
os.makedirs(mongo_db)
print('Starting embedded mongo daemon..')
sock = socket.socket()
sock.bind(('127.0.0.1', 0))
port = sock.getsockname()[1]
sock.close()
cls.mongo_db_uri = "mongodb://127.0.0.1:%s" % port
mongo_args = ['/usr/bin/mongod', '--dbpath', mongo_db, '--port',
str(port), '--logpath', mongo_log, '--smallfiles']
mp = cls._mongo_proc = subprocess.Popen(
mongo_args, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, shell=False)
print('Giving it some secs to correctly start..')
time_hacker.set_real_time()
# mongo takes some time to startup as it creates freshly new database files
# so we need a relatively big timeout:
timeout = time.time() + cls.mongod_start_timeout
while time.time() < timeout:
time.sleep(1)
mp.poll()
if mp.returncode is not None:
cls._read_mongolog_and_raise(mongo_log, mp,
"Launched mongod but it's directly died")
sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
errno = sock.connect_ex(('127.0.0.1', port))
if not errno:
sock.close()
break
else:
mp.kill()
cls._read_mongolog_and_raise(
mongo_log, mp,
"could not connect to port %s : mongod failed to correctly start?" % port)
time_hacker.set_my_time()
def test_a_long_history(self):
#return
test_host_005 = self.sched.hosts.find_by_name("test_host_005")
test_host_099 = self.sched.hosts.find_by_name("test_host_099")
test_ok_00 = self.sched.services.find_srv_by_name_and_hostname(
"test_host_005", "test_ok_00")
test_ok_01 = self.sched.services.find_srv_by_name_and_hostname(
"test_host_005", "test_ok_01")
test_ok_04 = self.sched.services.find_srv_by_name_and_hostname(
"test_host_005", "test_ok_04")
test_ok_16 = self.sched.services.find_srv_by_name_and_hostname(
"test_host_005", "test_ok_16")
test_ok_99 = self.sched.services.find_srv_by_name_and_hostname(
"test_host_099", "test_ok_01")
days = 4
etime = time.time()
print "now it is", time.ctime(etime)
print "now it is", time.gmtime(etime)
etime_midnight = (etime - (etime % 86400)) + time.altzone
print "midnight was", time.ctime(etime_midnight)
print "midnight was", time.gmtime(etime_midnight)
query_start = etime_midnight - (days - 1) * 86400
query_end = etime_midnight
print "query_start", time.ctime(query_start)
print "query_end ", time.ctime(query_end)
# |----------|----------|----------|----------|----------|---x
# etime
# etime_midnight
# ---x------
# etime - 4 days
# |---
# query_start
#
# ............................................
# events in the log database ranging till now
#
# |________________________________|
# events which will be read from db
#
loops = int(86400 / 192)
time_hacker.time_warp(-1 * days * 86400)
print "warp back to", time.ctime(time.time())
# run silently
old_stdout = sys.stdout
sys.stdout = open(os.devnull, "w")
should_be = 0
for day in xrange(days):
sys.stderr.write("day %d now it is %s i run %d loops\n" %
(day, time.ctime(time.time()), loops))
self.scheduler_loop(2, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
])
self.update_broker()
#for i in xrange(3600 * 24 * 7):
for i in xrange(loops):
if i % 10000 == 0:
sys.stderr.write(str(i))
if i % 399 == 0:
self.scheduler_loop(3, [
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
[test_ok_04, 3, "UNKN"],
[test_ok_16, 1, "WARN"],
[test_ok_99, 2, "CRIT"],
])
if int(time.time()) >= query_start and int(
time.time()) <= query_end:
should_be += 3
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(62)
if i % 399 == 0:
self.scheduler_loop(1, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
])
if int(time.time()) >= query_start and int(
time.time()) <= query_end:
should_be += 1
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(2)
if i % 17 == 0:
self.scheduler_loop(3, [
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
])
time.sleep(62)
if i % 17 == 0:
self.scheduler_loop(1, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
])
time.sleep(2)
if i % 14 == 0:
self.scheduler_loop(3, [
[test_host_005, 2, "DOWN"],
])
if i % 12 == 0:
self.scheduler_loop(3, [
[test_host_099, 2, "DOWN"],
])
time.sleep(62)
if i % 14 == 0:
self.scheduler_loop(3, [
[test_host_005, 0, "UP"],
])
if i % 12 == 0:
self.scheduler_loop(3, [
[test_host_099, 0, "UP"],
])
time.sleep(2)
self.update_broker()
if i % 1000 == 0:
self.livestatus_broker.db.commit()
endtime = time.time()
self.livestatus_broker.db.commit()
sys.stderr.write("day %d end it is %s\n" %
(day, time.ctime(time.time())))
sys.stdout.close()
sys.stdout = old_stdout
self.livestatus_broker.db.commit_and_rotate_log_db()
numlogs = self.livestatus_broker.db.execute(
"SELECT COUNT(*) FROM logs")
print "numlogs is", numlogs
# now we have a lot of events
# find type = HOST ALERT for test_host_005
request = """GET log
Columns: class time type state host_name service_description plugin_output message options contact_name command_name state_type current_host_groups current_service_groups
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
Filter: host_name = test_host_099
Filter: service_description = test_ok_01
And: 5
OutputFormat: json"""
# switch back to realtime. we want to know how long it takes
time_hacker.set_real_time()
print request
print "query 1 --------------------------------------------------"
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(
request)
tac = time.time()
pyresponse = eval(response)
print "number of records with test_ok_01", len(pyresponse)
self.assertEqual(should_be, len(pyresponse))
# and now test Negate:
request = """GET log
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
Filter: host_name = test_host_099
Filter: service_description = test_ok_01
And: 2
Negate:
And: 2
OutputFormat: json"""
response, keepalive = self.livestatus_broker.livestatus.handle_request(
request)
print "got response with true instead of negate"
notpyresponse = eval(response)
print "number of records without test_ok_01", len(notpyresponse)
request = """GET log
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
OutputFormat: json"""
response, keepalive = self.livestatus_broker.livestatus.handle_request(
request)
allpyresponse = eval(response)
print "all records", len(allpyresponse)
self.assert_(
len(allpyresponse) == len(notpyresponse) + len(pyresponse))
# the numlogs above only counts records in the currently attached db
numlogs = self.livestatus_broker.db.execute(
"SELECT COUNT(*) FROM logs WHERE time >= %d AND time <= %d" %
(int(query_start), int(query_end)))
print "numlogs is", numlogs
time_hacker.set_my_time()
def test_a_long_history(self):
# copied from test_livestatus_cache
test_host_005 = self.sched.hosts.find_by_name("test_host_005")
test_host_099 = self.sched.hosts.find_by_name("test_host_099")
find = self.sched.services.find_srv_by_name_and_hostname
test_ok_00 = find("test_host_005", "test_ok_00")
test_ok_01 = find("test_host_005", "test_ok_01")
test_ok_04 = find("test_host_005", "test_ok_04")
test_ok_16 = find("test_host_005", "test_ok_16")
test_ok_99 = find("test_host_099", "test_ok_01")
days = 4
etime = time.time()
print("now it is", time.ctime(etime))
print("now it is", time.gmtime(etime))
etime_midnight = (etime - (etime % 86400)) + time.altzone
print("midnight was", time.ctime(etime_midnight))
print("midnight was", time.gmtime(etime_midnight))
query_start = etime_midnight - (days - 1) * 86400
query_end = etime_midnight
print("query_start", time.ctime(query_start))
print("query_end ", time.ctime(query_end))
# |----------|----------|----------|----------|----------|---x
# etime
# etime_midnight
# ---x------
# etime - 4 days
# |---
# query_start
#
# ............................................
# events in the log database ranging till now
#
# |________________________________|
# events which will be read from db
#
loops = int(86400 / 192)
time_hacker.time_warp(-1 * days * 86400)
print("warp back to", time.ctime(time.time()))
# run silently
old_stdout = sys.stdout
sys.stdout = open(os.devnull, "w")
should_be = 0
for day in xrange(days):
sys.stderr.write("day %d now it is %s i run %d loops\n" % (
day, time.ctime(time.time()), loops))
self.scheduler_loop(2, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
])
self.update_broker()
#for i in xrange(3600 * 24 * 7):
for i in xrange(loops):
if i % 10000 == 0:
sys.stderr.write(str(i))
if i % 399 == 0:
self.scheduler_loop(3, [
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
[test_ok_04, 3, "UNKN"],
[test_ok_16, 1, "WARN"],
[test_ok_99, 2, "CRIT"],
])
if query_start <= int(time.time()) <= query_end:
should_be += 3
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(62)
if i % 399 == 0:
self.scheduler_loop(1, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
])
if query_start <= int(time.time()) <= query_end:
should_be += 1
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(2)
if i % 9 == 0:
self.scheduler_loop(3, [
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
])
time.sleep(62)
if i % 9 == 0:
self.scheduler_loop(1, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
])
time.sleep(2)
if i % 9 == 0:
self.scheduler_loop(3, [
[test_host_005, 2, "DOWN"],
])
if i % 2 == 0:
self.scheduler_loop(3, [
[test_host_099, 2, "DOWN"],
])
time.sleep(62)
if i % 9 == 0:
self.scheduler_loop(3, [
[test_host_005, 0, "UP"],
])
if i % 2 == 0:
self.scheduler_loop(3, [
[test_host_099, 0, "UP"],
])
time.sleep(2)
self.update_broker()
if i % 1000 == 0:
self.livestatus_broker.db.commit()
endtime = time.time()
self.livestatus_broker.db.commit()
sys.stderr.write("day %d end it is %s\n" % (day, time.ctime(time.time())))
sys.stdout.close()
sys.stdout = old_stdout
self.livestatus_broker.db.commit_and_rotate_log_db()
name = 'testtest' + self.testid
numlogs = self.livestatus_broker.db.conn[name].logs.find().count()
print("numlogs is", numlogs)
# now we have a lot of events
# find type = HOST ALERT for test_host_005
columns = (
"class time type state host_name service_description plugin_output message options "
"contact_name command_name state_type current_host_groups current_service_groups"
)
request = """GET log
Columns: """ + columns + """
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
Filter: host_name = test_host_099
Filter: service_description = test_ok_01
And: 5
OutputFormat: json"""
# switch back to realtime. we want to know how long it takes
time_hacker.set_real_time()
print(request)
print("query 1 --------------------------------------------------")
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
pyresponse = eval(response)
print(response)
print("number of records with test_ok_01", len(pyresponse))
self.assert_(len(pyresponse) == should_be)
# and now test Negate:
request = """GET log
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
Filter: host_name = test_host_099
Filter: service_description = test_ok_01
And: 2
Negate:
And: 2
OutputFormat: json"""
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
notpyresponse = eval(response)
print("number of records without test_ok_01", len(notpyresponse))
request = """GET log
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
OutputFormat: json"""
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
allpyresponse = eval(response)
print("all records", len(allpyresponse))
self.assert_(len(allpyresponse) == len(notpyresponse) + len(pyresponse))
# Now a pure class check query
request = """GET log
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: class = 1
OutputFormat: json"""
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
allpyresponse = eval(response)
print("all records", len(allpyresponse))
self.assert_(len(allpyresponse) == len(notpyresponse) + len(pyresponse))
# now delete too old entries from the database (> 14days)
# that's the job of commit_and_rotate_log_db()
numlogs = self.livestatus_broker.db.conn[name].logs.find().count()
times = [x['time'] for x in self.livestatus_broker.db.conn[name].logs.find()]
self.assert_(times != [])
print("whole database", numlogs, min(times), max(times))
numlogs = self.livestatus_broker.db.conn[name].logs.find({
'$and': [
{'time': {'$gt': min(times)}},
{'time': {'$lte': max(times)}}
]}).count()
now = max(times)
daycount = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
for day in xrange(25):
one_day_earlier = now - 3600*24
numlogs = self.livestatus_broker.db.conn[name].logs.find({
'$and': [
{'time': {'$gt': one_day_earlier}},
{'time': {'$lte': now}}
]}).count()
daycount[day] = numlogs
print("day -%02d %d..%d - %d" % (day, one_day_earlier, now, numlogs))
now = one_day_earlier
self.livestatus_broker.db.commit_and_rotate_log_db()
now = max(times)
for day in xrange(25):
one_day_earlier = now - 3600*24
numlogs = self.livestatus_broker.db.conn[name].logs.find({
'$and': [
{'time': {'$gt': one_day_earlier}},
{'time': {'$lte': now}}
]}).count()
print("day -%02d %d..%d - %d" % (day, one_day_earlier, now, numlogs))
now = one_day_earlier
numlogs = self.livestatus_broker.db.conn[name].logs.find().count()
# simply an estimation. the cleanup-routine in the mongodb logstore
# cuts off the old data at midnight, but here in the test we have
# only accuracy of a day.
self.assert_(numlogs >= sum(daycount[:7]))
self.assert_(numlogs <= sum(daycount[:8]))
time_hacker.set_my_time()
def setUp(self):
self.setup_with_file('etc/shinken_1r_1h_1s.cfg')
time_hacker.set_real_time()
Comment.id = 1
self.testid = str(os.getpid() + random.randint(1, 1000))
def setUp(self):
self.setup_with_file('etc/shinken_1r_1h_1s.cfg')
time_hacker.set_real_time()
Comment.id = 1
self.testid = str(os.getpid() + random.randint(1, 1000))
# GNU Affero General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with Shinken. If not, see <http://www.gnu.org/licenses/>.
"""
Test the hot dependencies arbiter module.
"""
import os, time
from shinken_test import unittest, ShinkenTest, time_hacker
# Need to use the real time-functions as we are reading timestamps
# from the filesystem.
time_hacker.set_real_time()
from shinken.objects.module import Module
from shinken.modulesctx import modulesctx
hot_dependencies_arbiter = modulesctx.get_module('hot_dependencies')
Hot_dependencies_arbiter = hot_dependencies_arbiter.Hot_dependencies_arbiter
get_instance = hot_dependencies_arbiter.get_instance
modconf = Module()
modconf.module_name = "PickleRetention"
modconf.module_type = hot_dependencies_arbiter.properties['type']
modconf.modules = []
modconf.properties = hot_dependencies_arbiter.properties.copy()
def test_a_long_history(self):
# copied from test_livestatus_cache
test_host_005 = self.sched.hosts.find_by_name("test_host_005")
test_host_099 = self.sched.hosts.find_by_name("test_host_099")
find = self.sched.services.find_srv_by_name_and_hostname
test_ok_00 = find("test_host_005", "test_ok_00")
test_ok_01 = find("test_host_005", "test_ok_01")
test_ok_04 = find("test_host_005", "test_ok_04")
test_ok_16 = find("test_host_005", "test_ok_16")
test_ok_99 = find("test_host_099", "test_ok_01")
days = 4
etime = time.time()
print("now it is", time.ctime(etime))
print("now it is", time.gmtime(etime))
etime_midnight = (etime - (etime % 86400)) + time.altzone
print("midnight was", time.ctime(etime_midnight))
print("midnight was", time.gmtime(etime_midnight))
query_start = etime_midnight - (days - 1) * 86400
query_end = etime_midnight
print("query_start", time.ctime(query_start))
print("query_end ", time.ctime(query_end))
# |----------|----------|----------|----------|----------|---x
# etime
# etime_midnight
# ---x------
# etime - 4 days
# |---
# query_start
#
# ............................................
# events in the log database ranging till now
#
# |________________________________|
# events which will be read from db
#
loops = int(86400 / 192)
time_hacker.time_warp(-1 * days * 86400)
print("warp back to", time.ctime(time.time()))
# run silently
old_stdout = sys.stdout
sys.stdout = open(os.devnull, "w")
should_be = 0
for day in xrange(days):
sys.stderr.write("day %d now it is %s i run %d loops\n" % (day, time.ctime(time.time()), loops))
self.scheduler_loop(
2,
[
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
],
)
self.update_broker()
# for i in xrange(3600 * 24 * 7):
for i in xrange(loops):
if i % 10000 == 0:
sys.stderr.write(str(i))
if i % 399 == 0:
self.scheduler_loop(
3,
[
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
[test_ok_04, 3, "UNKN"],
[test_ok_16, 1, "WARN"],
[test_ok_99, 2, "CRIT"],
],
)
if query_start <= int(time.time()) <= query_end:
should_be += 3
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(62)
if i % 399 == 0:
self.scheduler_loop(
1,
[
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
],
)
if query_start <= int(time.time()) <= query_end:
should_be += 1
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(2)
if i % 9 == 0:
self.scheduler_loop(3, [[test_ok_00, 1, "WARN"], [test_ok_01, 2, "CRIT"]])
time.sleep(62)
if i % 9 == 0:
self.scheduler_loop(1, [[test_ok_00, 0, "OK"], [test_ok_01, 0, "OK"]])
time.sleep(2)
if i % 9 == 0:
self.scheduler_loop(3, [[test_host_005, 2, "DOWN"]])
if i % 2 == 0:
self.scheduler_loop(3, [[test_host_099, 2, "DOWN"]])
time.sleep(62)
if i % 9 == 0:
self.scheduler_loop(3, [[test_host_005, 0, "UP"]])
if i % 2 == 0:
self.scheduler_loop(3, [[test_host_099, 0, "UP"]])
time.sleep(2)
self.update_broker()
if i % 1000 == 0:
self.livestatus_broker.db.commit()
endtime = time.time()
self.livestatus_broker.db.commit()
sys.stderr.write("day %d end it is %s\n" % (day, time.ctime(time.time())))
sys.stdout.close()
sys.stdout = old_stdout
self.livestatus_broker.db.commit_and_rotate_log_db()
name = "testtest" + self.testid
numlogs = self.livestatus_broker.db.conn[name].logs.find().count()
print("numlogs is", numlogs)
# now we have a lot of events
# find type = HOST ALERT for test_host_005
columns = (
"class time type state host_name service_description plugin_output message options "
"contact_name command_name state_type current_host_groups current_service_groups"
)
request = (
"""GET log
Columns: """
+ columns
+ """
Filter: time >= """
+ str(int(query_start))
+ """
Filter: time <= """
+ str(int(query_end))
+ """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
Filter: host_name = test_host_099
Filter: service_description = test_ok_01
And: 5
OutputFormat: json"""
)
# switch back to realtime. we want to know how long it takes
time_hacker.set_real_time()
print(request)
print("query 1 --------------------------------------------------")
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
pyresponse = eval(response)
print(response)
print("number of records with test_ok_01", len(pyresponse))
self.assert_(len(pyresponse) == should_be)
# and now test Negate:
request = (
"""GET log
Filter: time >= """
+ str(int(query_start))
+ """
Filter: time <= """
+ str(int(query_end))
+ """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
Filter: host_name = test_host_099
Filter: service_description = test_ok_01
And: 2
Negate:
And: 2
OutputFormat: json"""
)
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
notpyresponse = eval(response)
print("number of records without test_ok_01", len(notpyresponse))
request = (
"""GET log
Filter: time >= """
+ str(int(query_start))
+ """
Filter: time <= """
+ str(int(query_end))
+ """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
OutputFormat: json"""
)
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
allpyresponse = eval(response)
print("all records", len(allpyresponse))
self.assert_(len(allpyresponse) == len(notpyresponse) + len(pyresponse))
# Now a pure class check query
request = (
"""GET log
Filter: time >= """
+ str(int(query_start))
+ """
Filter: time <= """
+ str(int(query_end))
+ """
Filter: class = 1
OutputFormat: json"""
)
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
allpyresponse = eval(response)
print("all records", len(allpyresponse))
self.assert_(len(allpyresponse) == len(notpyresponse) + len(pyresponse))
# now delete too old entries from the database (> 14days)
# that's the job of commit_and_rotate_log_db()
numlogs = self.livestatus_broker.db.conn[name].logs.find().count()
times = [x["time"] for x in self.livestatus_broker.db.conn[name].logs.find()]
self.assert_(times != [])
print("whole database", numlogs, min(times), max(times))
numlogs = (
self.livestatus_broker.db.conn[name]
.logs.find({"$and": [{"time": {"$gt": min(times)}}, {"time": {"$lte": max(times)}}]})
.count()
)
now = max(times)
daycount = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
for day in xrange(25):
one_day_earlier = now - 3600 * 24
numlogs = (
self.livestatus_broker.db.conn[name]
.logs.find({"$and": [{"time": {"$gt": one_day_earlier}}, {"time": {"$lte": now}}]})
.count()
)
daycount[day] = numlogs
print("day -%02d %d..%d - %d" % (day, one_day_earlier, now, numlogs))
now = one_day_earlier
self.livestatus_broker.db.commit_and_rotate_log_db()
now = max(times)
for day in xrange(25):
one_day_earlier = now - 3600 * 24
numlogs = (
self.livestatus_broker.db.conn[name]
.logs.find({"$and": [{"time": {"$gt": one_day_earlier}}, {"time": {"$lte": now}}]})
.count()
)
print("day -%02d %d..%d - %d" % (day, one_day_earlier, now, numlogs))
now = one_day_earlier
numlogs = self.livestatus_broker.db.conn[name].logs.find().count()
# simply an estimation. the cleanup-routine in the mongodb logstore
# cuts off the old data at midnight, but here in the test we have
# only accuracy of a day.
self.assert_(numlogs >= sum(daycount[:7]))
self.assert_(numlogs <= sum(daycount[:8]))
time_hacker.set_my_time()
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU Affero General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with Shinken. If not, see <http://www.gnu.org/licenses/>.
"""
Test the hot dependencies arbiter module.
"""
import os, time
from shinken_test import unittest, ShinkenTest, time_hacker
# Need to use the real time-functions as we are reading timestamps
# from the filesystem.
time_hacker.set_real_time()
from shinken.objects.module import Module
from shinken.modulesctx import modulesctx
hot_dependencies_arbiter = modulesctx.get_module('hot_dependencies')
Hot_dependencies_arbiter = hot_dependencies_arbiter.Hot_dependencies_arbiter
get_instance = hot_dependencies_arbiter.get_instance
modconf = Module()
modconf.module_name = "PickleRetention"
modconf.module_type = hot_dependencies_arbiter.properties['type']
modconf.modules = []
modconf.properties = hot_dependencies_arbiter.properties.copy()
def setUp(self):
self.setup_with_file('etc/shinken_1r_1h_1s.cfg')
time_hacker.set_real_time()
def setUp(self):
self.setup_with_file('etc/shinken_1r_1h_1s.cfg')
time_hacker.set_real_time()
def test_a_long_history(self):
#return
test_host_005 = self.sched.hosts.find_by_name("test_host_005")
test_host_099 = self.sched.hosts.find_by_name("test_host_099")
test_ok_00 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_00")
test_ok_01 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_01")
test_ok_04 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_04")
test_ok_16 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_16")
test_ok_99 = self.sched.services.find_srv_by_name_and_hostname("test_host_099", "test_ok_01")
days = 4
etime = time.time()
print "now it is", time.ctime(etime)
print "now it is", time.gmtime(etime)
etime_midnight = (etime - (etime % 86400)) + time.altzone
print "midnight was", time.ctime(etime_midnight)
print "midnight was", time.gmtime(etime_midnight)
query_start = etime_midnight - (days - 1) * 86400
query_end = etime_midnight
print "query_start", time.ctime(query_start)
print "query_end ", time.ctime(query_end)
# |----------|----------|----------|----------|----------|---x
# etime
# etime_midnight
# ---x------
# etime - 4 days
# |---
# query_start
#
# ............................................
# events in the log database ranging till now
#
# |________________________________|
# events which will be read from db
#
loops = int(86400 / 192)
time_hacker.time_warp(-1 * days * 86400)
print "warp back to", time.ctime(time.time())
# run silently
old_stdout = sys.stdout
sys.stdout = open(os.devnull, "w")
should_be = 0
for day in xrange(days):
sys.stderr.write("day %d now it is %s i run %d loops\n" % (day, time.ctime(time.time()), loops))
self.scheduler_loop(2, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
])
self.update_broker()
#for i in xrange(3600 * 24 * 7):
for i in xrange(loops):
if i % 10000 == 0:
sys.stderr.write(str(i))
if i % 399 == 0:
self.scheduler_loop(3, [
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
[test_ok_04, 3, "UNKN"],
[test_ok_16, 1, "WARN"],
[test_ok_99, 2, "CRIT"],
])
if int(time.time()) >= query_start and int(time.time()) <= query_end:
should_be += 3
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(62)
if i % 399 == 0:
self.scheduler_loop(1, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
])
if int(time.time()) >= query_start and int(time.time()) <= query_end:
should_be += 1
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(2)
if i % 17 == 0:
self.scheduler_loop(3, [
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
])
time.sleep(62)
if i % 17 == 0:
self.scheduler_loop(1, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
])
time.sleep(2)
if i % 14 == 0:
self.scheduler_loop(3, [
[test_host_005, 2, "DOWN"],
])
if i % 12 == 0:
self.scheduler_loop(3, [
[test_host_099, 2, "DOWN"],
])
time.sleep(62)
if i % 14 == 0:
self.scheduler_loop(3, [
[test_host_005, 0, "UP"],
])
if i % 12 == 0:
self.scheduler_loop(3, [
[test_host_099, 0, "UP"],
])
time.sleep(2)
self.update_broker()
if i % 1000 == 0:
self.livestatus_broker.db.commit()
endtime = time.time()
self.livestatus_broker.db.commit()
sys.stderr.write("day %d end it is %s\n" % (day, time.ctime(time.time())))
sys.stdout.close()
sys.stdout = old_stdout
self.livestatus_broker.db.commit_and_rotate_log_db()
numlogs = self.livestatus_broker.db.execute("SELECT COUNT(*) FROM logs")
print "numlogs is", numlogs
# now we have a lot of events
# find type = HOST ALERT for test_host_005
request = """GET log
Columns: class time type state host_name service_description plugin_output message options contact_name command_name state_type current_host_groups current_service_groups
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
Filter: host_name = test_host_099
Filter: service_description = test_ok_01
And: 5
OutputFormat: json"""
# switch back to realtime. we want to know how long it takes
time_hacker.set_real_time()
print request
print "query 1 --------------------------------------------------"
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
pyresponse = eval(response)
print "number of records with test_ok_01", len(pyresponse)
self.assertEqual(should_be, len(pyresponse))
# and now test Negate:
request = """GET log
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
Filter: host_name = test_host_099
Filter: service_description = test_ok_01
And: 2
Negate:
And: 2
OutputFormat: json"""
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
print "got response with true instead of negate"
notpyresponse = eval(response)
print "number of records without test_ok_01", len(notpyresponse)
request = """GET log
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
OutputFormat: json"""
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
allpyresponse = eval(response)
print "all records", len(allpyresponse)
self.assert_(len(allpyresponse) == len(notpyresponse) + len(pyresponse))
# the numlogs above only counts records in the currently attached db
numlogs = self.livestatus_broker.db.execute("SELECT COUNT(*) FROM logs WHERE time >= %d AND time <= %d" % (int(query_start), int(query_end)))
print "numlogs is", numlogs
time_hacker.set_my_time()
def test_a_long_history(self):
print("Now: %s" % datetime.datetime.now())
print("Today: %s" % datetime.datetime.today())
test_host_005 = self.sched.hosts.find_by_name("test_host_005")
test_host_099 = self.sched.hosts.find_by_name("test_host_099")
test_ok_00 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_00")
test_ok_01 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_01")
test_ok_04 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_04")
test_ok_16 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_16")
test_ok_99 = self.sched.services.find_srv_by_name_and_hostname("test_host_099", "test_ok_01")
days = 4
etime = time.time()
now = time.time()
print("************* Now:", now)
print("now it is %s" % time.ctime(etime))
print("now it is %s" % time.gmtime(etime))
etime_midnight = (etime - (etime % 86400)) + time.altzone
print("midnight was %s" % time.ctime(etime_midnight))
print("midnight was %s" % time.gmtime(etime_midnight))
query_start = etime_midnight - (days - 1) * 86400
query_end = etime_midnight
print("query_start %s" % time.ctime(query_start))
print("query_end %s" % time.ctime(query_end))
# |----------|----------|----------|----------|----------|---x
# etime
# etime_midnight
# ---x------
# etime - 4 days
# |---
# query_start
#
# ............................................
# events in the log database ranging till now
#
# |________________________________|
# events which will be read from db
#
loops = int(86400 / 192)
time_hacker.time_warp(-1 * days * 86400)
print("warp back to %s" % time.ctime(time.time()))
# run silently
old_stdout = sys.stdout
sys.stdout = open(os.devnull, "w")
should_be = 0
for day in xrange(days):
sys.stderr.write("day %d now it is %s i run %d loops\n" % (day, time.ctime(time.time()), loops))
self.scheduler_loop(2, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
])
self.update_broker()
for i in xrange(loops):
if i % 10000 == 0:
sys.stderr.write(str(i))
if i % 399 == 0:
self.scheduler_loop(3, [
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
[test_ok_04, 3, "UNKN"],
[test_ok_16, 1, "WARN"],
[test_ok_99, 2, "CRIT"],
])
if int(time.time()) >= query_start and int(time.time()) <= query_end:
should_be += 3
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(62)
if i % 399 == 0:
self.scheduler_loop(1, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
])
if int(time.time()) >= query_start and int(time.time()) <= query_end:
should_be += 1
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(2)
if i % 9 == 0:
self.scheduler_loop(3, [
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
])
time.sleep(62)
if i % 9 == 0:
self.scheduler_loop(1, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
])
time.sleep(2)
if i % 9 == 0:
self.scheduler_loop(3, [
[test_host_005, 2, "DOWN"],
])
if i % 2 == 0:
self.scheduler_loop(3, [
[test_host_099, 2, "DOWN"],
])
time.sleep(62)
if i % 9 == 0:
self.scheduler_loop(3, [
[test_host_005, 0, "UP"],
])
if i % 2 == 0:
self.scheduler_loop(3, [
[test_host_099, 0, "UP"],
])
time.sleep(2)
self.update_broker()
if i % 1000 == 0:
self.livestatus_broker.db.commit()
endtime = time.time()
self.livestatus_broker.db.commit()
sys.stderr.write("day %d end it is %s\n" % (day, time.ctime(time.time())))
sys.stdout.close()
sys.stdout = old_stdout
self.livestatus_broker.db.commit_and_rotate_log_db()
print("************* Now + :", time.time() - now)
numlogs = self.livestatus_broker.db.execute("SELECT count(*) FROM logs")
print("Logs # is %s" % numlogs)
# now we have a lot of events
# find type = HOST ALERT for test_host_005
request = """GET log
Columns: class time type state host_name service_description plugin_output message options contact_name command_name state_type current_host_groups current_service_groups
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
Filter: host_name = test_host_099
Filter: service_description = test_ok_01
And: 5
OutputFormat: json"""
# switch back to realtime. we want to know how long it takes
time_hacker.set_real_time()
print("DB: %s" % self.livestatus_broker.db.database_file)
print("Request %s" % request)
print("query 1 cache---------------------------------------------")
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
elapsed1 = tac - tic
pyresponse = eval(response)
print("pyresponse %s" % len(pyresponse))
print("should be %s" % should_be)
self.assertEqual(should_be, len(pyresponse))
print("query 2 cache---------------------------------------------")
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
elapsed2 = tac - tic
pyresponse = eval(response)
self.assertEqual(should_be, len(pyresponse) )
print("clear the cache")
print("use aggressive sql")
print("query 3 --------------------------------------------------")
self.livestatus_broker.query_cache.wipeout()
self.livestatus_broker.db.use_aggressive_sql = True
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
elapsed3 = tac - tic
pyresponse = eval(response)
self.assertEqual(should_be, len(pyresponse))
print("query 4 cache---------------------------------------------")
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
elapsed4 = tac - tic
pyresponse = eval(response)
self.assertEqual(should_be, len(pyresponse))
print("Elapsed time:")
print("- elapsed1", elapsed1)
print("- elapsed2", elapsed2)
print("- elapsed3", elapsed3)
print("- elapsed4", elapsed4)
msg = """~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
NB: This isn't necessarily a failure !!!
This check highly depends on the system load while the test was running.
Maybe you could relaunch the test and it will succeed.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"""
self.assertLess(elapsed2*0.9, elapsed1, msg)
self.assertLess(elapsed3*0.9, elapsed1, msg)
self.assertLess(elapsed4*0.9, elapsed3, msg)
time_hacker.set_my_time()
def test_a_long_history(self):
#return
print datetime.datetime.now()
print datetime.datetime.today()
test_host_005 = self.sched.hosts.find_by_name("test_host_005")
test_host_099 = self.sched.hosts.find_by_name("test_host_099")
test_ok_00 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_00")
test_ok_01 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_01")
test_ok_04 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_04")
test_ok_16 = self.sched.services.find_srv_by_name_and_hostname("test_host_005", "test_ok_16")
test_ok_99 = self.sched.services.find_srv_by_name_and_hostname("test_host_099", "test_ok_01")
days = 4
etime = time.time()
print "now it is", time.ctime(etime)
print "now it is", time.gmtime(etime)
etime_midnight = (etime - (etime % 86400)) + time.altzone
print "midnight was", time.ctime(etime_midnight)
print "midnight was", time.gmtime(etime_midnight)
query_start = etime_midnight - (days - 1) * 86400
query_end = etime_midnight
print "query_start", time.ctime(query_start)
print "query_end ", time.ctime(query_end)
# |----------|----------|----------|----------|----------|---x
# etime
# etime_midnight
# ---x------
# etime - 4 days
# |---
# query_start
#
# ............................................
# events in the log database ranging till now
#
# |________________________________|
# events which will be read from db
#
loops = int(86400 / 192)
time_hacker.time_warp(-1 * days * 86400)
print "warp back to", time.ctime(time.time())
# run silently
old_stdout = sys.stdout
sys.stdout = open(os.devnull, "w")
should_be = 0
for day in xrange(days):
sys.stderr.write("day %d now it is %s i run %d loops\n" % (day, time.ctime(time.time()), loops))
self.scheduler_loop(2, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
])
self.update_broker()
#for i in xrange(3600 * 24 * 7):
for i in xrange(loops):
if i % 10000 == 0:
sys.stderr.write(str(i))
if i % 399 == 0:
self.scheduler_loop(3, [
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
[test_ok_04, 3, "UNKN"],
[test_ok_16, 1, "WARN"],
[test_ok_99, 2, "CRIT"],
])
if int(time.time()) >= query_start and int(time.time()) <= query_end:
should_be += 3
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(62)
if i % 399 == 0:
self.scheduler_loop(1, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
[test_ok_04, 0, "OK"],
[test_ok_16, 0, "OK"],
[test_ok_99, 0, "OK"],
])
if int(time.time()) >= query_start and int(time.time()) <= query_end:
should_be += 1
sys.stderr.write("now it should be %s\n" % should_be)
time.sleep(2)
if i % 9 == 0:
self.scheduler_loop(3, [
[test_ok_00, 1, "WARN"],
[test_ok_01, 2, "CRIT"],
])
time.sleep(62)
if i % 9 == 0:
self.scheduler_loop(1, [
[test_ok_00, 0, "OK"],
[test_ok_01, 0, "OK"],
])
time.sleep(2)
if i % 9 == 0:
self.scheduler_loop(3, [
[test_host_005, 2, "DOWN"],
])
if i % 2 == 0:
self.scheduler_loop(3, [
[test_host_099, 2, "DOWN"],
])
time.sleep(62)
if i % 9 == 0:
self.scheduler_loop(3, [
[test_host_005, 0, "UP"],
])
if i % 2 == 0:
self.scheduler_loop(3, [
[test_host_099, 0, "UP"],
])
time.sleep(2)
self.update_broker()
if i % 1000 == 0:
self.livestatus_broker.db.commit()
endtime = time.time()
self.livestatus_broker.db.commit()
sys.stderr.write("day %d end it is %s\n" % (day, time.ctime(time.time())))
sys.stdout.close()
sys.stdout = old_stdout
self.livestatus_broker.db.commit_and_rotate_log_db()
numlogs = self.livestatus_broker.db.execute("SELECT count(*) FROM logs")
print "numlogs is", numlogs
# now we have a lot of events
# find type = HOST ALERT for test_host_005
request = """GET log
Columns: class time type state host_name service_description plugin_output message options contact_name command_name state_type current_host_groups current_service_groups
Filter: time >= """ + str(int(query_start)) + """
Filter: time <= """ + str(int(query_end)) + """
Filter: type = SERVICE ALERT
And: 1
Filter: type = HOST ALERT
And: 1
Filter: type = SERVICE FLAPPING ALERT
Filter: type = HOST FLAPPING ALERT
Filter: type = SERVICE DOWNTIME ALERT
Filter: type = HOST DOWNTIME ALERT
Filter: type ~ starting...
Filter: type ~ shutting down...
Or: 8
Filter: host_name = test_host_099
Filter: service_description = test_ok_01
And: 5
OutputFormat: json"""
# switch back to realtime. we want to know how long it takes
time_hacker.set_real_time()
print self.livestatus_broker.db.database_file
print request
print "query 1 --------------------------------------------------"
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
elapsed1 = tac - tic
pyresponse = eval(response)
print "pyresponse", len(pyresponse)
print "should be", should_be
self.assertEqual(should_be, len(pyresponse))
print "query 2 cache---------------------------------------------"
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
elapsed2 = tac - tic
pyresponse = eval(response)
self.assertEqual(should_be, len(pyresponse) )
print "clear the cache"
print "use aggressive sql"
print "query 3 --------------------------------------------------"
self.livestatus_broker.query_cache.wipeout()
self.livestatus_broker.db.use_aggressive_sql = True
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
elapsed3 = tac - tic
pyresponse = eval(response)
self.assertEqual(should_be, len(pyresponse))
print "query 4 cache---------------------------------------------"
tic = time.time()
response, keepalive = self.livestatus_broker.livestatus.handle_request(request)
tac = time.time()
elapsed4 = tac - tic
pyresponse = eval(response)
self.assertEqual(should_be, len(pyresponse))
print "elapsed1", elapsed1
print "elapsed2", elapsed2
print "elapsed3", elapsed3
print "elapsed4", elapsed4
msg = """~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
NB NB NB: This isn't necessarily a failure !!! This check highly depends on the system load there was while the test was running.
Maybe you could relaunch the test and it will succeed.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"""
self.assertLess(elapsed2*0.9, elapsed1, msg)
self.assertLess(elapsed3*0.9, elapsed1, msg)
self.assertLess(elapsed4*0.9, elapsed3, msg)
time_hacker.set_my_time()