forked from redhat-performance/benchmark-runner
-
Notifications
You must be signed in to change notification settings - Fork 0
/
vdbench_pod.py
165 lines (157 loc) · 8.74 KB
/
vdbench_pod.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
import os
import time
from multiprocessing import Process
from benchmark_runner.common.logger.logger_time_stamp import logger_time_stamp, logger
from benchmark_runner.common.elasticsearch.elasticsearch_exceptions import ElasticSearchDataNotUploaded
from benchmark_runner.workloads.workloads_operations import WorkloadsOperations
from benchmark_runner.common.prometheus.prometheus_metrics_operations import PrometheusMetricsOperation
class VdbenchPod(WorkloadsOperations):
"""
This class run vdbench pod
"""
def __init__(self):
super().__init__()
self.__name = ''
self.__workload_name = ''
self.__es_index = ''
self.__kind = ''
self.__status = ''
self.__pod_name = ''
self.__scale = ''
self.__data_dict = {}
self.__prometheus_metrics_operation = PrometheusMetricsOperation()
def __create_pod_scale(self, pod_num: str):
"""
This method create pod in parallel
"""
self._oc.create_async(yaml=os.path.join(f'{self._run_artifacts_path}', f'{self.__name}_{pod_num}.yaml'))
self._oc.wait_for_pod_create(pod_name=f'{self.__pod_name}-{pod_num}')
def __run_pod_scale(self, pod_num: str):
"""
This method runs pod in parallel
"""
self._oc.wait_for_initialized(label=f'app=vdbench-{self._trunc_uuid}-{pod_num}', label_uuid=False)
self._oc.wait_for_ready(label=f'app=vdbench-{self._trunc_uuid}-{pod_num}', label_uuid=False)
self.__status = self._oc.wait_for_pod_completed(label=f'app=vdbench-{self._trunc_uuid}-{pod_num}', label_uuid=False, job=False)
self.__status = 'complete' if self.__status else 'failed'
self.__prometheus_metrics_operation.finalize_prometheus()
metric_results = self.__prometheus_metrics_operation.run_prometheus_queries()
prometheus_result = self.parse_prometheus_metrics(data=metric_results)
# save run artifacts logs
result_list = self._create_pod_run_artifacts(pod_name=f'{self.__pod_name}-{pod_num}', log_type='.csv')
if self._es_host:
# upload several run results
for result in result_list:
result.update(prometheus_result)
self._upload_to_elasticsearch(index=self.__es_index, kind=self.__kind, status=self.__status, result=result)
# verify that data upload to elastic search according to unique uuid
self._verify_elasticsearch_data_uploaded(index=self.__es_index, uuid=self._uuid)
def __delete_pod_scale(self, pod_num: str):
"""
This method create pod in parallel
"""
self._oc.delete_async(yaml=os.path.join(f'{self._run_artifacts_path}', f'{self.__name}_{pod_num}.yaml'))
@logger_time_stamp
def run(self):
"""
This method run the workload
:return:
"""
try:
self.__prometheus_metrics_operation.init_prometheus()
if 'kata' in self._workload:
self.__kind = 'kata'
self.__name = self._workload.replace('kata', 'pod')
else:
self.__kind = 'pod'
self.__name = self._workload
self.__workload_name = self._workload.replace('_', '-')
self.__pod_name = f'{self.__workload_name}-{self._trunc_uuid}'
if self._run_type == 'test_ci':
self.__es_index = 'vdbench-test-ci-results'
else:
self.__es_index = 'vdbench-results'
self._environment_variables_dict['kind'] = self.__kind
# create namespace
self._oc.create_async(yaml=os.path.join(f'{self._run_artifacts_path}', 'namespace.yaml'))
self._oc.apply_security_privileged()
if self.__kind == 'kata':
self._oc.set_kata_threads_pool(thread_pool_size=self._kata_thread_pool_size)
if not self._scale:
self._oc.create_pod_sync(yaml=os.path.join(f'{self._run_artifacts_path}', f'{self.__name}.yaml'), pod_name=self.__pod_name)
self._oc.wait_for_initialized(label=f'app=vdbench-{self._trunc_uuid}', label_uuid=False)
self._oc.wait_for_ready(label=f'app=vdbench-{self._trunc_uuid}', label_uuid=False)
self.__status = self._oc.wait_for_pod_completed(label=f'app=vdbench-{self._trunc_uuid}', label_uuid=False, job=False)
self.__status = 'complete' if self.__status else 'failed'
self.__prometheus_metrics_operation.finalize_prometheus()
metric_results = self.__prometheus_metrics_operation.run_prometheus_queries()
prometheus_result = self.parse_prometheus_metrics(data=metric_results)
# save run artifacts logs
result_list = self._create_pod_run_artifacts(pod_name=self.__pod_name, log_type='.csv')
if self._es_host:
# upload several run results
for result in result_list:
result.update(prometheus_result)
self._upload_to_elasticsearch(index=self.__es_index, kind=self.__kind, status=self.__status, result=result)
# verify that data upload to elastic search according to unique uuid
self._verify_elasticsearch_data_uploaded(index=self.__es_index, uuid=self._uuid)
self._oc.delete_pod_sync(
yaml=os.path.join(f'{self._run_artifacts_path}', f'{self.__name}.yaml'),
pod_name=self.__pod_name)
# scale
else:
self.__scale = int(self._scale)
# create redis and state signals
sync_pods = {'redis': 'redis', 'state_signals_exporter_pod': 'state-signals-exporter'}
for pod, name in sync_pods.items():
if pod == 'redis':
pod_name = f'redis-master'
else:
pod_name = name
self._oc.create_pod_sync(yaml=os.path.join(f'{self._run_artifacts_path}', f'{pod}.yaml'), pod_name=pod_name)
self._oc.wait_for_initialized(label=f'app={name}', label_uuid=False)
self._oc.wait_for_ready(label=f'app={name}', label_uuid=False)
# prepare scale run
bulks = tuple(self.split_run_bulks(iterable=range(self._scale * len(self._scale_node_list)), limit=self._threads_limit))
# create, run and delete vms
for target in (self.__create_pod_scale, self.__run_pod_scale, self.__delete_pod_scale):
proc = []
for bulk in bulks:
for pod_num in bulk:
p = Process(target=target, args=(str(pod_num),))
p.start()
proc.append(p)
for p in proc:
p.join()
# sleep between bulks
time.sleep(self._bulk_sleep_time)
proc = []
self._create_scale_logs()
# delete redis and state signals
for pod, name in sync_pods.items():
if pod == 'redis':
pod_name = f'redis-master'
else:
pod_name = name
self._oc.delete_pod_sync(yaml=os.path.join(f'{self._run_artifacts_path}', f'{pod}.yaml'), pod_name=pod_name)
# delete namespace
self._oc.delete_async(yaml=os.path.join(f'{self._run_artifacts_path}', 'namespace.yaml'))
if self.__kind == 'kata':
self._oc.delete_kata_threads_pool()
except ElasticSearchDataNotUploaded as err:
self._oc.delete_pod_sync(
yaml=os.path.join(f'{self._run_artifacts_path}', f'{self.__name}.yaml'),
pod_name=self.__pod_name)
raise err
except Exception as err:
# save run artifacts logs
if self._oc.pod_exists(pod_name=self.__pod_name):
self._create_pod_log(pod=self.__pod_name)
self.__data_dict['run_artifacts_url'] = os.path.join(self._run_artifacts_url, f'{self._get_run_artifacts_hierarchy(workload_name=self.__workload_name, is_file=True)}-{self._time_stamp_format}.tar.gz')
if self._es_host:
self._upload_to_elasticsearch(index=self.__es_index, kind=self.__kind, status='failed', result=self.__data_dict)
# verify that data upload to elastic search according to unique uuid
self._verify_elasticsearch_data_uploaded(index=self.__es_index, uuid=self._uuid)
self._oc.delete_pod_sync(
yaml=os.path.join(f'{self._run_artifacts_path}', f'{self.__name}.yaml'), pod_name=self.__pod_name)
raise err