blob: 4cd94ba55cd677c972fa68b069c757e5ad750c7f [file] [log] [blame]
Dan Shif8b71d12014-09-03 10:31:08 -07001#!/usr/bin/env python
2
3# Copyright (c) 2014 The Chromium OS Authors. All rights reserved.
4# Use of this source code is governed by a BSD-style license that can be
5# found in the LICENSE file.
6
7# Script to check the history of stage calls made to devserver.
8# Following are some sample use cases:
9#
10# 1. Find all stage request for autotest and image nyan_big-release/R38-6055.0.0
11# in the last 10 days across all devservers.
12# ./devserver_history.py --image_filters nyan_big 38 6055.0.0 -l 240 \
13# --artifact_filters autotest -v
14# output:
15# ==============================================================================
16# 170.21.64.22
17# ==============================================================================
18# Number of calls: 1
19# Number of unique images: 1
20# 2014-08-23 12:45:00: nyan_big-release/R38-6055.0.0 autotest
21# ==============================================================================
22# 170.21.64.23
23# ==============================================================================
24# Number of calls: 2
25# Number of unique images: 1
26# 2014-08-23 12:45:00: nyan_big-release/R38-6055.0.0 autotest, test_suites
27# 2014-08-23 12:55:00: nyan_big-release/R38-6055.0.0 autotest, test_suites
28#
29# 2. Find all duplicated stage request for the last 10 days.
30# ./devserver_history.py -d -l 240
31# output:
32# Detecting artifacts staged in multiple devservers.
33# ==============================================================================
34# nyan_big-release/R38-6055.0.0
35# ==============================================================================
36# 170.21.64.22: 23 requests 2014-09-04 22:44:28 -- 2014-09-05 00:03:23
37# 170.21.64.23: 6 requests 2014-09-04 22:48:58 -- 2014-09-04 22:49:42
38#
39# Count of images with duplicated stages on each devserver:
40# 170.21.64.22 : 22
41# 170.21.64.23 : 11
42
43
44import argparse
45import datetime
46import logging
47import operator
48import re
49import time
50from itertools import groupby
51
52import common
53from autotest_lib.client.common_lib import global_config
54from autotest_lib.client.common_lib import time_utils
55from autotest_lib.client.common_lib.cros.graphite import es_utils
56
57
58class devserver_call(object):
59 """A container to store the information of devserver stage call.
60 """
61
62 def __init__(self, hit):
63 """Retrieve information from a ES query hit.
64 """
65 self.devserver = hit['_source']['devserver']
66 self.subname = hit['_source']['subname']
67 self.artifacts = hit['_source']['artifacts'].split(' ')
68 self.image = hit['_source']['image']
69 self.value = hit['_source']['value']
70 self.time_recorded = time_utils.epoch_time_to_date_string(
71 hit['_source']['time_recorded'])
72
73
74 def __str__(self):
75 pairs = ['%-20s: %s' % (attr, getattr(self, attr)) for attr in dir(self)
76 if not attr.startswith('__') and
77 not callable(getattr(self, attr))]
78 return '\n'.join(pairs)
79
80
81def get_calls(time_start, time_end, artifact_filters=None,
82 regex_constraints=None, devserver=None, size=1e7):
83 """Gets all devserver calls from es db with the given constraints.
84
85 @param time_start: Earliest time entry was recorded.
86 @param time_end: Latest time entry was recorded.
87 @param artifact_filters: A list of names to match artifacts.
88 @param regex_constraints: A list of regex constraints for ES query.
89 @param devserver: name of devserver to query for. If it's set to None,
90 return calls for all devservers. Default is set to None.
91 @param size: Max number of entries to return, default to 1 million.
92
93 @returns: Entries from esdb.
94 """
95 eqs = [('_type', 'devserver')]
96 if devserver:
97 eqs.append(('devserver', devserver))
98 if artifact_filters:
99 for artifact in artifact_filters:
100 eqs.append(('artifacts', artifact))
101 query = es_utils.create_range_eq_query_multiple(
102 fields_returned=None,
103 equality_constraints=eqs,
104 range_constraints=[('time_recorded', time_start, time_end)],
105 size=size,
106 sort_specs=[{'time_recorded': 'desc'}],
107 regex_constraints=regex_constraints)
108 results = es_utils.execute_query(query, timeout=30)
109 devserver_calls = []
110 for hit in results['hits']['hits']:
111 devserver_calls.append(devserver_call(hit))
112 logging.info('Found %d calls.', len(devserver_calls))
113 return devserver_calls
114
115
116def print_call_details(calls, verbose):
117 """Print details of each call to devserver to stage artifacts.
118
119 @param calls: A list of devserver stage requests.
120 @param verbose: Set to True to print out all devserver calls.
121 """
122 calls = sorted(calls, key=lambda c: c.devserver)
123 for devserver,calls_for_devserver in groupby(calls, lambda c: c.devserver):
124 calls_for_devserver = list(calls_for_devserver)
125 print '='*80
126 print devserver
127 print '='*80
128 print 'Number of calls: %d' % len(calls_for_devserver)
129 print ('Number of unique images: %d' %
130 len(set([call.image for call in calls_for_devserver])))
131 if verbose:
132 for call in sorted(calls_for_devserver,
133 key=lambda c: c.time_recorded):
134 print ('%s %s %s' % (call.time_recorded, call.image,
135 ', '.join(call.artifacts)))
136
137
138def detect_duplicated_stage(calls):
139 """Detect any artifact for same build was staged in multiple devservers.
140
141 @param calls: A list of devserver stage requests.
142 """
143 print '\nDetecting artifacts staged in multiple devservers.'
144 calls = sorted(calls, key=lambda c: c.image)
145 # Count how many times a devserver staged duplicated artifacts. A number
146 # significantly larger then others can indicate that the devserver failed
147 # check_health too often and needs to be removed from production.
148 duplicated_stage_count = {}
149 for image,calls_for_image in groupby(calls, lambda c: c.image):
150 calls_for_image = list(calls_for_image)
151 devservers = set([call.devserver for call in calls_for_image])
152 if len(devservers) > 1:
153 print '='*80
154 print image
155 print '='*80
156 calls_for_image = sorted(calls_for_image, key=lambda c: c.devserver)
157 for devserver,calls_for_devserver in groupby(calls_for_image,
158 lambda c: c.devserver):
159 timestamps = [c.time_recorded for c in calls_for_devserver]
160 print ('%s: %-3d requests %s -- %s' %
161 (devserver, len(timestamps), min(timestamps),
162 max(timestamps)))
163 duplicated_stage_count[devserver] = (
164 duplicated_stage_count.get(devserver, 0) + 1)
165 print '\nCount of images with duplicated stages on each devserver:'
166 counts = sorted(duplicated_stage_count.iteritems(),
167 key=operator.itemgetter(1), reverse=True)
168 for k,v in counts:
169 print '%-15s: %d' % (k, v)
170
171
172def main():
173 """main script. """
174 t_now = time.time()
175 t_now_minus_one_day = t_now - 3600 * 24
176 parser = argparse.ArgumentParser()
177 parser.add_argument('-l', type=float, dest='last',
178 help='last hours to search results across',
179 default=None)
180 parser.add_argument('--start', type=str, dest='start',
181 help=('Enter start time as: yyyy-mm-dd hh-mm-ss,'
182 'defualts to 24h ago. This option is ignored when'
183 ' -l is used.'),
184 default=time_utils.epoch_time_to_date_string(
185 t_now_minus_one_day))
186 parser.add_argument('--end', type=str, dest='end',
187 help=('Enter end time in as: yyyy-mm-dd hh-mm-ss,'
188 'defualts to current time. This option is ignored'
189 ' when -l is used.'),
190 default=time_utils.epoch_time_to_date_string(t_now))
191 parser.add_argument('--devservers', nargs='+', dest='devservers',
192 help=('Enter space deliminated devservers. Default are'
193 ' all devservers specified in global config.'),
194 default=[])
195 parser.add_argument('--artifact_filters', nargs='+',
196 dest='artifact_filters',
197 help=('Enter space deliminated filters on artifact '
198 'name. For example "autotest test_suites". The '
199 'filter does not support regex.'),
200 default=[])
201 parser.add_argument('--image_filters', nargs='+', dest='image_filters',
202 help=('Enter space deliminated filters on image name. '
203 'For example "nyan 38 6566", search will use '
204 'regex to match each filter. Do not use filters '
205 'with mixed letter and number, e.g., R38.'),
206 default=[])
207 parser.add_argument('-d', '--detect_duplicated_stage', action='store_true',
208 dest='detect_duplicated_stage',
209 help=('Set to True to detect if an artifacts for a same'
210 ' build was staged in multiple devservers. '
211 'Default is True.'),
212 default=False)
213 parser.add_argument('-v', action='store_true', dest='verbose',
214 default=False,
215 help='-v to print out ALL entries.')
216 options = parser.parse_args()
217 if options.verbose:
218 logging.getLogger().setLevel(logging.INFO)
219
220 if options.last:
221 end_time = datetime.datetime.now()
222 start_time = end_time - datetime.timedelta(seconds=3600 * options.last)
223 else:
224 start_time = datetime.datetime.strptime(options.start,
225 time_utils.TIME_FMT)
226 end_time = datetime.datetime.strptime(options.end, time_utils.TIME_FMT)
227 logging.info('Searching devserver calls from %s to %s', start_time,
228 end_time)
229
230 devservers = options.devservers
231 if not devservers:
232 devserver_urls = global_config.global_config.get_config_value(
233 'CROS', 'dev_server', type=list, default=[])
234 devservers = []
235 for url in devserver_urls:
236 match = re.match('http://([^:]*):*\d*', url)
237 devservers.append(match.groups(0)[0] if match else url)
238 logging.info('Found devservers: %s', devservers)
239
240 regex_constraints = []
241 for filter in options.image_filters:
242 regex_constraints.append(('image', '.*%s.*' % filter))
243 calls = []
244 for devserver in devservers:
245 calls.extend(get_calls(start_time, end_time, options.artifact_filters,
246 regex_constraints, devserver=devserver))
247
248 print_call_details(calls, options.verbose)
249
250 if options.detect_duplicated_stage:
251 detect_duplicated_stage(calls)
252
253
254if __name__ == '__main__':
255 main()