forked from gioenn/dynaSpark-launcher
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathaggregate_analysis.py
941 lines (858 loc) · 47.4 KB
/
aggregate_analysis.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
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
from spark_time_analysis import run as run_ta
from spark_log_profiling import processing as profiling
import glob
import os
import sys
import plotly.plotly as py
import plotly.graph_objs as go
from functools import reduce
import numpy as np
import argparse
import spark_time_analysis.cfg as ta_cfg
import json
import util.utils as utils
import shutil
import pprint
import collections
import math
import config
from util.ssh_client import CustomSSHClient
from plumbum.machines.paramiko_machine import ParamikoMachine
from plumbum import BG, FG
from util.plot_analyses import get_scatter, get_scatter2, get_layout, plot_figure
from time import sleep
from concurrent.futures import ThreadPoolExecutor
import random
MAX_WORKERS = 4
DEFAULT_SEARCH_ORDER = {'uppaal': 'breadth-first',
'zot': 'depth-first'}
SEARCH_ORDERS = ['breadth-first', 'depth-first']
REMOTE_D_VERT_SERVERS = {
'azure': {
'hostname': '70.37.53.100',
'username': 'ubuntu',
'json2mc_path': os.path.abspath(os.path.join(os.sep, 'home', 'ubuntu', 'DICE-Verification',
'd-vert-server', 'd-vert-json2mc')),
'exp_dir': os.path.abspath(os.path.join(os.sep, 'home', 'ubuntu', 'FAC_2019_exp')),
'env_path': [''],
},
'azure_fac18_0': {
'hostname': '104.215.95.166',
'username': 'ubuntu',
'json2mc_path': os.path.abspath(os.path.join(os.sep, 'home', 'ubuntu', 'DICE-Verification',
'd-vert-server', 'd-vert-json2mc')),
'exp_dir': os.path.abspath(os.path.join(os.sep, 'home', 'ubuntu', 'FAC_2019_exp')),
'env_path': [''],
},
'azure_fac19_0': {
'hostname': '104.214.50.119',
'username': 'ubuntu',
'json2mc_path': os.path.abspath(os.path.join(os.sep, 'home', 'ubuntu', 'DICE-Verification',
'd-vert-server', 'd-vert-json2mc')),
'exp_dir': os.path.abspath(os.path.join(os.sep, 'home', 'ubuntu', 'FAC_2019_exp')),
'env_path': [''],
},
'azure_fac19_1': {
'hostname': '104.214.56.40',
'username': 'ubuntu',
'json2mc_path': os.path.abspath(os.path.join(os.sep, 'home', 'ubuntu', 'DICE-Verification',
'd-vert-server', 'd-vert-json2mc')),
'exp_dir': os.path.abspath(os.path.join(os.sep, 'home', 'ubuntu', 'FAC_2019_exp')),
'env_path': [''],
},
'fm_biased': {
'hostname': 'planetlab1.elet.polimi.it',
'username': 'fmbiased',
'json2mc_path': os.path.abspath(os.path.join(os.sep, 'home', 'fmbiased', 'DICE', 'Francesco', 'd4s',
'd-vert-server', 'd-vert-json2mc')),
'exp_dir': os.path.abspath(os.path.join(os.sep, 'home', 'fmbiased', 'FAC_2019_exp')),
'env_path': ['/home/fmbiased/DICE/Francesco/zot/bin:/home/fmbiased/DICE/Francesco/z3/bin:'
'/home/fmbiased/uppaal64-4.1.19/bin-Linux'],
},
'marce_biased': {
'hostname': 'planetlab1.elet.polimi.it',
'username': 'bersani',
'json2mc_path': os.path.abspath(os.path.join(os.sep, 'home', 'fmbiased', 'DICE', 'Francesco', 'd4s',
'd-vert-server', 'd-vert-json2mc')),
'exp_dir': os.path.abspath(os.path.join(os.sep, 'home', 'bersani', 'FAC_2019_exp')),
'env_path': [''],
},
}
SERVER_NAMES = list(REMOTE_D_VERT_SERVERS.keys())
EXP_DIR = os.path.join('d4s_FAC')
ABS_EXP_DIR = {
'marce_biased': os.path.abspath(os.path.join(os.sep, 'home', 'bersani', 'FAC_2019_exp')),
'fm_biased': os.path.abspath(os.path.join(os.sep, 'home', 'bersani', 'FAC_2019_exp')),
'azure': os.path.abspath(os.path.join(os.sep, 'home', 'ubuntu', 'FAC_2019_exp'))
}
PATH = {
'fm_biased': ['/home/fmbiased/DICE/Francesco/zot/bin:/home/fmbiased/DICE/Francesco/z3/bin:/home/fmbiased/uppaal64-4.1.19/bin-Linux'],
'marce_biased': ['/home/fmbiased/DICE/Francesco/zot/bin:/home/fmbiased/DICE/Francesco/z3/bin:/home/fmbiased/uppaal64-4.1.19/bin-Linux'],
'azure': ['/usr/local/uppaal64-4.1.19/bin-Linux/'],
}
DEFAULT_NUM_RECORDS = 200000000
DEFAULT_NUM_CORES = 16
IMGS_FOLDER = 'imgs'
CONTEXTS_FOLDER = 'contexts'
ESSENTIAL_FILES = ['app.json',
'app.dat',
'config.json',
'*_time_analysis.json',]
JOB_STATS = ['actual_job_duration',
'total_ta_executor_stages',
'total_ta_master_stages',
'total_overhead_monocore',
'GQ_master'] + ['total_percentile' + str(p) for p in run_ta.PERCENTILES]
STAGES_STATS = ['io_factor',
's_avg_duration_ta_master',
's_avg_duration_ta_executor',
's_GQ_ta_master',
't_avg_duration_ta_executor',
't_avg_duration_ta_master',
't_mean_no_init',
't_record_ta_master',
't_std_dev',]
JOB_STATS_BIG_JSON = ['actual_job_duration',
'num_cores',
'num_of_points',
'num_v',]
STAGES_STATS_BIG_JSON = ['add_to_end_taskset',
'actual_records_read',
'io_factor',
's_GQ_ta_master',
's_GQ_ta_executor',
't_record_ta_executor',
't_record_ta_master',
't_task_ta_master',
'task_durations',]
SIMPLE_AVERAGE_STATS = ['avg_actual_job_duration',
'avg_total_ta_executor_stages',
'avg_total_ta_master_stages'] + ['avg_total_percentile' + str(p) for p in run_ta.PERCENTILES]
COMBINED_STATS = ['avg_total_with_avg_gq_and_ta_master',
'avg_total_with_avg_gq_and_ta_executor',
'avg_total_with_avg_gq_and_ta_executor_plus_overhead',
'avg_total_with_local_gq_and_ta_master',
'avg_total_with_avg_gq_and_avg_t_record_master',
'avg_total_with_avg_gq_and_local_t_record_master',]
PLOT_EXEC_TIMES_STATS = SIMPLE_AVERAGE_STATS + COMBINED_STATS
# t_task selection policies
GQ_AVG_T_REC_AVG = 't_task'
GQ_AVG_T_REC_LOC = 't_task_num_v'
SIGMA_0_25 = 't_task_0_25_sigma'
SIGMA_0_30 = 't_task_0_30_sigma'
SIGMA_0_80 = 't_task_0_80_sigma'
NUM_RECORDS_FACTOR = {
'pagerank': 20,
'kmeans': 2,
'sort_by_key': 20000000,
'louvain': 1,
}
PERCENTAGES = [
# 0,
1,
# 2,
# 3,
5,
# 7,
10,
15,
20,
# # -1,
# # -5,
# -10,
# -20,
]
def gen_perc(arg, coeff):
return int(arg + arg*coeff/100)
def get_num_records(bench, param):
return param * NUM_RECORDS_FACTOR[bench]
def get_records_read(stages_struct, num_records, modify_stages_struct=False):
"""
computes the number of records read/write for all the stages and modifies stages_struct to include it.
:param stages_struct: data structure containing the
:param num_records: total number of input records
:param modify_stages_struct: enable modification of stages_struct by inserting the computed number f records read
:returns reads dictionary
"""
reads = {}
writes = {}
stage_id_list = [int(x) for x in stages_struct.keys()]
stage_id_list.sort()
for i in stage_id_list:
stage = stages_struct[str(i)]
stage_id = str(i)
if len(stage['parentsIds']) == 0:
# print(stage_id)
if not num_records:
num_records = stage['actual_records_read']
reads[stage_id] = num_records
else:
reads[stage_id] = 0
# print(stage_id)
for parent_id in stage['parentsIds']:
reads[stage_id] += writes[str(parent_id)]
writes[stage_id] = reads[stage_id] * stage['avg_io_factor']
if modify_stages_struct:
stage['records_read'] = reads[stage_id]
return reads
def compute_t_task(stages_struct, num_records, num_cores, benchmark, num_task=None, t_task_policy=GQ_AVG_T_REC_AVG):
"""
computes t_task for all the stages and modifies stages_struct to include it.
:param benchmark: benchmark application name
:param stages_struct: data structure containing the stats regarding all the stages
:param num_records: total number of input records
:param num_cores: number of cores in the cluster
:param num_task: number of tasks for each stages (currently uniform)
:param t_task_policy: policy to select the t_task for verification
(that will be stored in stage['t_task_verification'])
:returns t_tasks dictionary, t_tasks_num_v dictionary, num_tasks dictionary
"""
get_records_read(stages_struct, num_records, True)
for k, stage in stages_struct.items():
if not num_task:
num_task = stage['numtask']
# compute t_task with avg_t_record and avg_gq
stage['t_task'] = stage['avg_t_record'] * stage['records_read'] / (num_task * stage['avg_gq'])
stage['t_task_local'] = {}
stage['t_task_avg'] = {}
stage['t_task_0_25_sigma_local'] = {}
stage['t_task_0_30_sigma_local'] = {}
stage['t_task_0_80_sigma_local'] = {}
num_batches = math.ceil(num_task / num_cores)
# TODO remove this approximation (it only works when rounded_tasks == num_tasks)
rounded_tasks = num_cores * num_batches
for v in stage['avg_t_record_num_v'].keys():
# compute t_task with "local" avg_t_record_num_v and avg_gq
tmp_reads = get_records_read(stages_struct, NUM_RECORDS_FACTOR[benchmark] * int(v))
stage['t_task_local'][v] = stage['avg_t_record_num_v'][v] * tmp_reads[k] / (rounded_tasks * stage['avg_gq'])
#stage['t_task_local'][v] = stage['avg_t_record_num_v'][v] * tmp_reads[k] / (num_task * stage['avg_gq'])
stage['t_task_avg'][v] = stage['avg_t_record'] * tmp_reads[k] / (rounded_tasks * stage['avg_gq'])
#stage['t_task_avg'][v] = stage['avg_t_record'] * tmp_reads[k] / (num_task * stage['avg_gq'])
stage['t_task_0_25_sigma_local'][v] = stage['avg_t_avg_duration_ta_master'][v] + 0.25 * \
stage['avg_t_std_dev'][v]
stage['t_task_0_30_sigma_local'][v] = stage['avg_t_avg_duration_ta_master'][v] + 0.3 * \
stage['avg_t_std_dev'][v]
stage['t_task_0_80_sigma_local'][v] = stage['avg_t_avg_duration_ta_master'][v] + 0.8 * \
stage['avg_t_std_dev'][v]
print('num_records: {}'.format(num_records))
num_v = str(int(num_records / NUM_RECORDS_FACTOR[benchmark]))
print('if {} in {}:'.format(num_v, stage['t_task_local']))
if num_v in stage['t_task_local']:
stage['t_task_num_v'] = stage['t_task_local'][num_v]
stage['t_task_0_25_sigma'] = stage['t_task_0_25_sigma_local'][num_v]
stage['t_task_0_30_sigma'] = stage['t_task_0_30_sigma_local'][num_v]
stage['t_task_0_80_sigma'] = stage['t_task_0_80_sigma_local'][num_v]
else:
stage['t_task_num_v'] = 0
stage['t_task_0_25_sigma'] = 0
stage['t_task_0_30_sigma'] = 0
stage['t_task_0_80_sigma'] = 0
stage['t_task_verification'] = stage[t_task_policy]
return {s['id']: s['t_task'] for s in stages_struct.values()}, \
{s['id']: s['t_task_num_v'] for s in stages_struct.values()}, \
{s['id']: num_task for s in stages_struct.values()}
def build_generic_stages_struct(profiled_stages, res): # avg_gq, avg_t_record, avg_io, avg_gq_num_v, avg_t_record_num_v):
generic_stages_struct = {}
for k, v in profiled_stages.items():
generic_stages_struct[k] = {}
generic_stages_struct[k]['id'] = v['id']
# generic_stages_struct[k]['name'] = v['name']
generic_stages_struct[k]['parentsIds'] = v['parent_ids']
generic_stages_struct[k]['skipped'] = v['skipped']
generic_stages_struct[k]['numtask'] = v['num_task']
generic_stages_struct[k]['avg_gq'] = np.mean(list(res['avg_s_GQ_ta_master'][k].values())) # avg_gq[k]
generic_stages_struct[k]['avg_gq_num_v'] = res['avg_s_GQ_ta_master'][k] # avg_gq_num_v[k]
generic_stages_struct[k]['avg_t_record'] = np.mean(list(res['avg_t_record_ta_master'][k].values())) #avg_t_record[k]
generic_stages_struct[k]['avg_t_record_num_v'] = res['avg_t_record_ta_master'][k] # avg_t_record_num_v[k]
generic_stages_struct[k]['avg_io_factor'] = np.mean(list(res['avg_io_factor'][k].values())) # avg_io[k]
generic_stages_struct[k]['avg_t_avg_duration_ta_master'] = res['avg_t_avg_duration_ta_master'][k]
generic_stages_struct[k]['avg_t_avg_duration_ta_executor'] = res['avg_t_avg_duration_ta_executor'][k]
generic_stages_struct[k]['avg_t_std_dev'] = res['avg_t_std_dev'][k]
return generic_stages_struct
def calculate_sequential_duration(generic_stages_struct, num_tasks, num_cores, t_task_policy):
seq_duration_avg_t_record = seq_duration_local_t_record = seq_duration_sigma_0_25 = selected_seq_duration = seq_duration_sigma_0_30 = seq_duration_sigma_0_80 = 0
for k, v in generic_stages_struct.items():
if not num_tasks:
num_tasks = v['numtask']
else:
v['numtask'] = num_tasks
num_batches = math.ceil(num_tasks / num_cores)
seq_duration_local_t_record += v['t_task_num_v'] * num_batches
seq_duration_avg_t_record += v['t_task'] * num_batches
seq_duration_sigma_0_25 += v['t_task_0_25_sigma'] * num_batches
seq_duration_sigma_0_30 += v['t_task_0_30_sigma'] * num_batches
seq_duration_sigma_0_80 += v['t_task_0_80_sigma'] * num_batches
selected_seq_duration += v[t_task_policy] * num_batches
print('S{}\t-> tmp "local" sequential duration: {}ms\t(+{})'.format(k, int(seq_duration_local_t_record),
v['t_task_num_v'] * num_batches))
print('S{}\t-> tmp average sequential duration: {}ms\t(+{})'.format(k, int(seq_duration_avg_t_record),
v['t_task'] * num_batches))
print('S{}\t-> tmp 0_25_sigma sequential duration: {}ms\t(+{})'.format(k, int(seq_duration_sigma_0_25),
v['t_task_0_25_sigma'] * num_batches))
print('S{}\t-> tmp 0_30_sigma sequential duration: {}ms\t(+{})'.format(k, int(seq_duration_sigma_0_30),
v['t_task_0_30_sigma'] * num_batches))
print('S{}\t-> tmp 0_80_sigma sequential duration: {}ms\t(+{})'.format(k, int(seq_duration_sigma_0_80),
v['t_task_0_80_sigma'] * num_batches))
print('estimated "local" sequential duration: {}ms'.format(int(seq_duration_local_t_record)))
print('estimated average sequential duration: {}ms'.format(int(seq_duration_avg_t_record)))
print('estimated 0_25_sigma sequential duration: {}ms'.format(int(seq_duration_sigma_0_25)))
print('estimated 0_30_sigma sequential duration: {}ms'.format(int(seq_duration_sigma_0_30)))
print('estimated 0_80_sigma sequential duration: {}ms'.format(int(seq_duration_sigma_0_80)))
return selected_seq_duration
def generate_spark_context(args):
exp_dir = os.path.abspath(args.exp_dir)
run_verification = args.verify
analysis_id = exp_dir.strip('/').split('/')[-1]
num_records = args.num_records if args.num_records else DEFAULT_NUM_RECORDS
num_cores = args.num_cores if args.num_cores else DEFAULT_NUM_CORES
deadlines = args.deadlines
num_tasks = args.num_tasks
time_bound = args.time_bound if args.time_bound else [ta_cfg.TIME_BOUND]
server = args.server
engine = args.engine
max_workers = args.max_workers if args.max_workers else MAX_WORKERS
labeling = args.labeling
d_range = args.d_range
d_percentages = args.d_percentages
search_order = args.search_order if args.search_order else DEFAULT_SEARCH_ORDER[engine]
plugin = args.plugin if args.plugin else ta_cfg.PLUGIN
print('generate_spark_context for num_records: {}'.format(num_records))
aggregated_stats_path = glob.glob(os.path.join(exp_dir, '{}_aggregated_stats.json'.format(analysis_id)))
generic_stages_path = glob.glob(os.path.join(exp_dir, '{}_generic_stages.json'.format(analysis_id)))
if not generic_stages_path or not aggregated_stats_path:
print('{}_generic_stages.json FILE NOT FOUND!\nRUN PROFILING/TIME_ANALYSIS FIRST'.format(analysis_id))
sys.exit(1)
else:
with open(generic_stages_path[0]) as gsf:
generic_stages_struct = json.load(gsf)
print('opening {}'.format(aggregated_stats_path[0]))
with open(aggregated_stats_path[0]) as asf:
aggregated_stats = json.load(asf)
benchmark = aggregated_stats['benchmark_name']
t_task_policy = SIGMA_0_80
compute_t_task(stages_struct=generic_stages_struct, num_records=num_records, num_task=num_tasks,
t_task_policy=t_task_policy, num_cores=num_cores, benchmark=benchmark)
selected_seq_duration = calculate_sequential_duration(generic_stages_struct=generic_stages_struct,
num_tasks=num_tasks, num_cores=num_cores,
t_task_policy=t_task_policy)
if not deadlines:
deadlines = [int(selected_seq_duration)]
elif d_range: # expects 3 values for deadlines
epsilon = (1, -1)[deadlines[2] < 0] # allows for second term inclusion
deadlines = list(range(deadlines[0], deadlines[1] + epsilon, deadlines[2]))
elif d_percentages:
deadlines = [gen_perc(deadlines[0], y) for y in PERCENTAGES]
contexts_dir = os.path.join(exp_dir, CONTEXTS_FOLDER)
context_files_struct = {}
for tb in time_bound:
for d in deadlines:
print('Generating JSON file for deadline {}, time_bound: {}'.format(d, tb))
app_name = "{}_c{}_t{}_nr{}_tb{}_{}l_d{}" \
"_tc_{}_n_rounds_{}_{}_{}_{}".format(analysis_id,
num_cores,
num_tasks,
num_records,
tb,
"no_" if ta_cfg.NO_LOOPS else "",
d,
"parametric" if ta_cfg.PARAMETRIC_TC else
'{}_{}'.format(num_cores,
num_cores -
num_tasks % num_cores),
"by1", t_task_policy,
search_order,
"label" if labeling else "no_label")
# "exp_dir_acceleration_0_1000_c48_t40_no-l_d133000_tc_parametric_forall_nrounds_TEST",
SPARK_CONTEXT = {
"app_name": app_name,
"app_type": benchmark,
"verification_params":
{
"plugin": plugin,
"time_bound": tb,
"parametric_tc": ta_cfg.PARAMETRIC_TC,
"no_loops": ta_cfg.NO_LOOPS
},
"tot_cores": num_cores,
"analysis_type": "feasibility",
"deadline": d,
"max_time": d,
"tolerance": ta_cfg.TOLERANCE,
"stages": generic_stages_struct,
"labeling": True if labeling else False,
"search_order": search_order,
}
utils.make_sure_path_exists(contexts_dir)
out_path_context = os.path.join(contexts_dir, '{}_context.json'.format(app_name))
print("dumping to {}".format(out_path_context))
with open(out_path_context, 'w') as outfile:
json.dump(SPARK_CONTEXT, outfile, indent=4, sort_keys=True)
context_files_struct['{}__{}'.format(tb, d)] = out_path_context
if run_verification:
od = collections.OrderedDict(sorted(context_files_struct.items(), reverse=True))
with ThreadPoolExecutor(max_workers) as executor:
for k, v in od.items():
print("TIMEBOUND__DEADLINE: {}\nFile: {}".format(k, v))
executor.submit(ssh_launch_json2mc, v, server, engine, labeling)
def launch_verification(args):
json_path = args.json
tasks = args.num_tasks
labeling = args.labeling
server = args.server
engine = args.engine
max_workers = args.max_workers if args.max_workers else MAX_WORKERS
run_verification = args.verify
search_order = args.search_order if args.search_order else DEFAULT_SEARCH_ORDER[engine]
d_range = args.d_range
d_percentages = args.d_percentages
plugin = args.plugin if args.plugin else ta_cfg.PLUGIN
with open(json_path) as cf:
context = json.load(cf)
input_records = context['stages']['0']['records_read'] if 'records_read' in context['stages']['0'] \
else context['stages']['0']['recordsread']
deadlines = args.deadlines if args.deadlines else [context['deadline']]
print("DEADLINES: {}".format(deadlines))
time_bound = args.time_bound if args.time_bound else [context['verification_params']['time_bound']]
context['tot_cores'] = args.num_cores if args.num_cores else context['tot_cores']
context['search_order'] = search_order
context['verification_params']['plugin'] = plugin
skipped_stages = []
if tasks:
for k,v in context["stages"].items():
v['numtask'] = tasks
for k, v in context["stages"].items():
if 't_task_verification' not in v:
if 'duration' in v:
v['t_task_verification'] = v['duration']/v['numtask']
else:
skipped_stages.append(k)
for s in skipped_stages:
context["stages"].pop(s)
contexts_dir = os.path.join(os.path.dirname(json_path), 'generated_contexts')
context_files_struct = {}
# range_end = deadlines[0]
# reverse_deadlines_list = list(reversed(range(range_end - 10, range_end, 1)))
if not deadlines:
deadlines = [int(selected_seq_duration)]
elif d_range: # expects 3 values for deadlines
epsilon = (1, -1)[deadlines[2] < 0] # allows for second term inclusion
deadlines = list(range(deadlines[0], deadlines[1] + epsilon, deadlines[2]))
elif d_percentages:
deadlines = [gen_perc(deadlines[0], y) for y in PERCENTAGES]
for tb in time_bound:
context['verification_params']['time_bound'] = tb
# for d in reverse_deadlines_list:
for d in deadlines:
app_name = '{}_c{}_t{}_nr{}_tb{}_d{}_{}_{}'.format(context['app_type'], context['tot_cores'],
tasks if tasks else 'default', input_records, tb, d,
search_order,'label' if labeling else 'NO_label')
context['app_name'] = app_name
context['deadline'] = context['max_time'] = d
out_path_context = os.path.join(contexts_dir, '{}_context.json'.format(app_name))
utils.make_sure_path_exists(contexts_dir)
print("dumping to {}".format(out_path_context))
with open(out_path_context, 'w') as outfile:
json.dump(context, outfile, indent=4, sort_keys=True)
context_files_struct['{}__{}'.format(tb, d)] = out_path_context
if run_verification:
od = collections.OrderedDict(sorted(context_files_struct.items(), reverse=True))
with ThreadPoolExecutor(max_workers) as executor:
for k, v in od.items():
print("TIMEBOUND__DEADLINE: {}\nFile: {}".format(k, v))
executor.submit(ssh_launch_json2mc, v, server, engine, labeling)
def generate_plots(res, stages_keys, input_dir, num_v_set, benchmark):
x_axis_int = [int(v) for v in num_v_set]
x_axis_int.sort()
x_axis = [str(v) for v in x_axis_int]
print('X_AXIS: {}'.format(x_axis))
trace_list = [get_scatter2(x_axis, res, stat) for stat in PLOT_EXEC_TIMES_STATS]
data_exec_times = go.Data(trace_list)
trace_list_avg_gq = []
trace_list_std_gq = []
trace_list_avg_t_record = []
trace_list_std_t_record = []
for k in stages_keys:
trace_list_avg_gq.append(get_scatter2(x_axis, res['avg_s_GQ_ta_master'], str(k)))
trace_list_std_gq.append(get_scatter2(x_axis, res['std_s_GQ_ta_master'], str(k)))
trace_list_avg_t_record.append(get_scatter2(x_axis, res['avg_t_record_ta_master'], str(k)))
trace_list_std_t_record.append(get_scatter2(x_axis, res['std_t_record_ta_master'], str(k)))
data_gq_stages = go.Data(trace_list_avg_gq)
data_t_record_stages = go.Data(trace_list_avg_t_record)
plot_figure(data=data_gq_stages,
title='average_GQ_{}'.format(input_dir.strip('/').split('/')[-1]),
x_axis_label="Num Vertices",
y_axis_label='Value ([0, 1])',
out_folder=os.path.join(input_dir, IMGS_FOLDER))
plot_figure(data=data_t_record_stages,
title='average_record_time_{}'.format(input_dir.strip('/').split('/')[-1]),
x_axis_label="Num Vertices",
y_axis_label='Time (ms)',
out_folder=os.path.join(input_dir, IMGS_FOLDER))
plot_figure(data=data_exec_times,
title='{}_execution_times_{}'.format(benchmark, input_dir.strip('/').split('/')[-1]),
x_axis_label="Num Vertices",
y_axis_label='Time (ms)',
out_folder=os.path.join(input_dir, IMGS_FOLDER))
def extract_essential_files(input_dir):
analysis_files_dir = os.path.abspath(os.path.join(os.path.dirname(input_dir.strip(os.sep)),
'ta_only',
'{}_time_analysis'.format(
input_dir.strip(os.sep).split(os.sep)[-1])))
print('analysis_files_dir: {}'.format(analysis_files_dir))
utils.make_sure_path_exists(analysis_files_dir)
for d in glob.glob(os.path.join(input_dir, 'app-*')):
dest_dir = os.path.join(analysis_files_dir, d.split(os.sep)[-1])
utils.make_sure_path_exists(dest_dir)
for f in ESSENTIAL_FILES:
for x in glob.glob(os.path.join(d, f)):
print('copying:\t{}\nto:\t{}'.format(x, dest_dir))
shutil.copy(x, dest_dir)
def collect_all_time_analysis(exp_dir):
input_dir = os.path.abspath(exp_dir)
out_path = os.path.join(input_dir, '{}_allinone_stats.json'.format(input_dir.split(os.sep)[-1]))
print("Getting time_analysis data from all the experiments in {}".format(input_dir))
res = {"directory": input_dir, "experiments": []}
for d in glob.glob(os.path.join(input_dir, 'app-*')):
print(d)
for t in glob.glob(os.path.join(d, '*_time_analysis.json')):
with open(t) as ta_file:
cur_ta = json.load(ta_file)
tmp_exp_report = {'job': {}, 'stages': {}}
tmp_exp_report['job']['id'] = os.path.basename(t)
for x in JOB_STATS_BIG_JSON:
try:
tmp_exp_report['job'][x] = cur_ta['job'][x]
except KeyError as e:
print("Key not found: {}".format(e))
for k, v in cur_ta['stages'].items():
tmp_exp_report['stages'][k] = {}
for x in STAGES_STATS_BIG_JSON:
tmp_exp_report['stages'][k][x] = v[x]
res['experiments'].append(tmp_exp_report)
print("dumping to {}".format(out_path))
with open(out_path, 'w') as outfile:
json.dump(res, outfile, indent=4, sort_keys=True)
def get_empty_dict_of_dicts(keys):
return {k: collections.defaultdict(list) for k in keys}
def time_analysis(args):
"""
launch time_analysis on selected_folder
:param args: expected arguments: exp_dir, plot, reprocess, collect_all_ta, extract_essentials
"""
# get command line arguments
input_dir = args.exp_dir
plot = args.plot
reprocess = args.reprocess
collect_all_ta = args.collect_all_ta
extract_essentials = args.extract_essentials
# executors = args.executors
analysis_id = input_dir.strip('/').split('/')[-1]
num_v_set = set([])
stages_sample = job_sample = None
exp_report2 = {} # exp-report2[STAGE/JOB][NUM_V]
ta_master = ta_master_avg = None
for x in JOB_STATS:
exp_report2[x] = collections.defaultdict(list)
# check for different directory structure (spark-bench or spark-perf)
app_dirs_spark_bench = glob.glob(os.path.join(input_dir, 'app-*'))
app_dirs_spark_perf = glob.glob(os.path.join(input_dir, 'spark_perf_output_*', 'app-*'))
app_dirs = app_dirs_spark_bench if app_dirs_spark_bench else app_dirs_spark_perf
# iterate over all the application directories included in input_dir
for d in app_dirs:
'''
if executors: # if specified, modify max_executor in config.json --> to be removed
run_ta.modify_executors(d, executors)
'''
if reprocess: # run time_analysis on d
ta_job, ta_stages = run_ta.main(d)
else: # get precomputed analysis file from d
ta_file_paths = glob.glob(os.path.join(d, '*_time_analysis.json'))
if ta_file_paths:
print("getting time_analysis from {}...".format(ta_file_paths[0]))
with open(ta_file_paths[0]) as ta_file:
ta_total = json.load(ta_file)
ta_job = ta_total['job']
ta_stages = ta_total['stages']
else: # if precomputed analysis is not available, launch time_analysis on current directory d
ta_job, ta_stages = run_ta.main(d)
# save numV from configuration files of current directory
benchmark = ta_job['benchmark_name']
par_var_name = config.VAR_PAR_MAP[benchmark]['var_name']
par_var = ta_job[par_var_name][1] if isinstance(ta_job[par_var_name], list) else ta_job[par_var_name]
num_v = str(par_var)
num_v_set.add(num_v)
if not stages_sample: # initialize all the data structures that will be used to store statistics
for x in STAGES_STATS:
exp_report2[x] = get_empty_dict_of_dicts(ta_stages.keys())
stages_sample = ta_stages
job_sample = ta_job
for x in JOB_STATS:
exp_report2[x][num_v].append(ta_job[x])
for k in ta_stages.keys():
for x in STAGES_STATS:
exp_report2[x][k][num_v].append(ta_stages[k][x])
exp_report2['benchmark_name'] = benchmark
if collect_all_ta:
collect_all_time_analysis(input_dir)
if extract_essentials:
extract_essential_files(input_dir)
resulting_stats = {}
# compute average and standard deviation of all the statistics
for k in JOB_STATS:
resulting_stats['avg_{}'.format(k)] = {}
resulting_stats['std_{}'.format(k)] = {}
for v in num_v_set:
resulting_stats['avg_{}'.format(k)][v] = np.mean(list(exp_report2[k][v]))
resulting_stats['std_{}'.format(k)][v] = np.std(list(exp_report2[k][v]))
for s in STAGES_STATS:
resulting_stats['avg_{}'.format(s)] = get_empty_dict_of_dicts(ta_stages.keys())
resulting_stats['std_{}'.format(s)] = get_empty_dict_of_dicts(ta_stages.keys())
for k in ta_stages.keys():
for v in num_v_set:
resulting_stats['avg_{}'.format(s)][k][v] = np.mean(list(exp_report2[s][k][v]))
resulting_stats['std_{}'.format(s)][k][v] = np.std(list(exp_report2[s][k][v]))
resulting_stats['benchmark_name'] = benchmark
out_path_exp_rep = os.path.join(input_dir, '{}_collected_stats.json'.format(analysis_id))
print("dumping collected_stats to {}".format(out_path_exp_rep))
with open(out_path_exp_rep, 'w+') as outfile:
json.dump(exp_report2, outfile, indent=4, sort_keys=True)
out_path_res = os.path.join(input_dir, '{}_aggregated_stats.json'.format(analysis_id))
print("dumping aggregated_stats to {}".format(out_path_res))
with open(out_path_res, 'w+') as outfile:
json.dump(resulting_stats, outfile, indent=4, sort_keys=True)
# build generic stages dict including all the average values for stats
generic_stages_dict = build_generic_stages_struct(profiled_stages=stages_sample, res=resulting_stats)
out_path_generic_s = os.path.join(input_dir, '{}_generic_stages.json'.format(analysis_id))
print("dumping generic_stages to {}".format(out_path_generic_s))
with open(out_path_generic_s, 'w+') as outfile:
json.dump(generic_stages_dict, outfile, indent=4, sort_keys=True)
# build estimates with different combinations
t_tasks = {}
t_tasks_num_v = {}
num_tasks = {}
print("num_v_set: {}\nnum_cores: {}".format(num_v_set, job_sample['num_cores']))
for x in COMBINED_STATS:
resulting_stats[x] = {}
for v in num_v_set:
resulting_stats[x][v] = 0
num_cores = job_sample['num_cores']
for v in num_v_set:
t_tasks[v], t_tasks_num_v[v], num_tasks[v] = compute_t_task(stages_struct=generic_stages_dict,
num_records=int(v) * NUM_RECORDS_FACTOR[benchmark],
num_cores=num_cores, benchmark=benchmark)
for s in ta_stages.keys():
ta_master = resulting_stats['avg_s_avg_duration_ta_master'][s][v]
avg_gq = generic_stages_dict[s]['avg_gq']
avg_gq_num_v = generic_stages_dict[s]['avg_gq_num_v'][v]
ta_executor = resulting_stats['avg_s_avg_duration_ta_executor'][s][v]
num_batches = math.ceil(num_tasks[v][s] / num_cores)
resulting_stats['avg_total_with_avg_gq_and_ta_master'][v] += ta_master / avg_gq
resulting_stats['avg_total_with_avg_gq_and_ta_executor'][v] += ta_executor / avg_gq
resulting_stats['avg_total_with_local_gq_and_ta_master'][v] += ta_master / avg_gq_num_v
resulting_stats['avg_total_with_avg_gq_and_avg_t_record_master'][v] += t_tasks[v][s] * num_batches
resulting_stats['avg_total_with_avg_gq_and_local_t_record_master'][v] += t_tasks_num_v[v][s] * num_batches
resulting_stats['avg_total_with_avg_gq_and_ta_executor_plus_overhead'][v] = \
resulting_stats['avg_total_with_avg_gq_and_ta_executor'][v] + \
resulting_stats['avg_total_overhead_monocore'][v] / num_cores
pp = pprint.PrettyPrinter(indent=4)
# pp.pprint(exp_report2)
if plot:
# generate_plots(res, ta_stages.keys(), input_dir)
generate_plots(resulting_stats, ta_stages.keys(), input_dir, num_v_set, benchmark)
def pro_runner(args):
reprocess = args.reprocess
exp_dir = args.exp_dir
app_dirs_spark_bench = glob.glob(os.path.join(exp_dir, 'app-*'))
app_dirs_spark_perf = glob.glob(os.path.join(exp_dir, 'spark_perf_output_*', 'app-*'))
app_dirs = app_dirs_spark_bench if app_dirs_spark_bench else app_dirs_spark_perf
for d in app_dirs:
profiling.main(input_dir=d, json_out_dir=d, reprocess=reprocess)
'''
def ssh_conn(args):
"""
apparently is not possible to run json2mc in background with only paramiko
:param args:
:return:
"""
filepath = args.file_path
client = CustomSSHClient(hostname=D_VERT_SERVER_HOSTNAME,
port=22,
username='ubuntu',
password=None,
key_files=config.PRIVATE_KEY_PATH)
client.connect()
destination_path = os.path.join(BASE_JSON2MC_PATH, 'd4s', os.path.basename(filepath))
client.put(localpath=filepath,
remotepath=destination_path)
client.run('. {}/venv/bin/activate'.format(BASE_JSON2MC_PATH))
status, std_out, std_err = client.run('cd {} && ./run_json2mc.py -T spark --db -c {} \&'.format(BASE_JSON2MC_PATH, destination_path))
print('std_err: {}\nstd_out {}\nstatus {}'.format(std_err, std_out, status))
'''
def ssh_launch_json2mc(filepath, server, engine, labeling):
"""
simple method that uploads the file whose path is provided as argument filepath
and remotely launches a verification task in background
:param filepath: path of the .json which has to be uploaded on the server and provided as a parameter to json2mc.py
"""
d_vert_server_hostname = REMOTE_D_VERT_SERVERS[server]['hostname']
base_json2mc_path = REMOTE_D_VERT_SERVERS[server]['json2mc_path']
username = REMOTE_D_VERT_SERVERS[server]['username']
print('ssh_launch_json2mc({})'.format(filepath))
# out_path = os.path.join(base_json2mc_path, EXP_DIR)
# destination_path = os.path.join(out_path, os.path.basename(filepath))
out_path = REMOTE_D_VERT_SERVERS[server]['exp_dir']
destination_path = os.path.join(out_path, os.path.basename(filepath))
# log_path = os.path.join(BASE_JSON2MC_PATH, 'logs', '{}.log'.format(os.path.splitext(os.path.basename(filepath))[0]))
print('connecting to {}'.format(d_vert_server_hostname))
rem = ParamikoMachine(host=d_vert_server_hostname, keyfile=config.PRIVATE_KEY_PATH, user=username)
print("mkdir {}".format(out_path))
mkdir = rem['mkdir']
mkdir['-p', out_path]()
print("adding {} to $PATH".format(REMOTE_D_VERT_SERVERS[server]['env_path']))
rem.env.path.insert(0, REMOTE_D_VERT_SERVERS[server]['env_path'])
print('uploading\n{}\nto\n{}:{}'.format(filepath, d_vert_server_hostname, destination_path))
rem.upload(filepath, destination_path)
with rem.cwd(base_json2mc_path):
activate_venv = rem['./activate_venv.sh']
run_json2mc = rem['./run_json2mc.py']
# source = rem['source']
# workon = rem['workon']
# source['~/.bashrc']()
print('activating venv...')
activate_venv()
# workon['d-vert']()
print('launching json2mc...')
print(' '.join(['./run_json2mc.py','-T', 'spark', '-e', engine, '--db', '-l', '-c', destination_path, '-o', out_path]))
# run_json2mc['-T', 'spark', '--db', '-c', destination_path, '-o', out_path] & FG
sleep(random.uniform(0, 3))
if labeling: # TODO improve this
f = run_json2mc['-T', 'spark', '-e', engine, '--db', '-l', '-c', destination_path, '-o', out_path] & BG
else:
f = run_json2mc['-T', 'spark', '-e', engine, '--db', '-c', destination_path, '-o', out_path] & BG
#f = (run_json2mc['-T', 'spark', '--db', '-c', destination_path] > 'pine.log')() & BG
print('launched {}'.format(f))
sleep(1)
f.wait()
if f.ready():
print('Command exited with return_code {}\nSTDOUT:{}\nSTDERR:{}'.format(f.returncode, f.stdout, f.stderr))
else:
print('Command running in background...\n{}'.format(f))
if __name__ == "__main__":
parser = argparse.ArgumentParser(
description=
"""
Aggregated Analysis Tool for xSparkBench
"""
)
subparsers = parser.add_subparsers()
parser_pro = subparsers.add_parser('pro', help='launch profiling on selected folders')
parser_ta = subparsers.add_parser('ta', help='launch time_analysis on selected_folder')
parser_gen = subparsers.add_parser('gen', help='generate json file for formal analysis')
parser_ver = subparsers.add_parser('ver', help='directly run verification given a json file and some settings')
parser_pro.add_argument("exp_dir", help="directory containing all the experiment files to be analyzed")
parser_pro.add_argument("-r", "--reprocess", dest="reprocess", action="store_true",
help="reprocess data (look for logs in processed_logs folders)"
"[default: %(default)s]")
parser_ta.add_argument("exp_dir", help="directory containing all the experiment files to be analyzed")
parser_ta.add_argument("-r", "--reprocess", dest="reprocess", action="store_true",
help="reprocess data (look for logs in provided folders)"
"[default: %(default)s]")
parser_ta.add_argument("-p", "--plot", dest="plot", action="store_true",
help="plots the performed analyses"
"[default: %(default)s]")
parser_ta.add_argument("-c", "--collect", dest="collect_all_ta", action="store_true",
help="collect some of the main important statistics in one json file "
"[default: %(default)s]")
parser_ta.add_argument("-e", "--extract-essentials", dest="extract_essentials", action="store_true",
help='extract essential files to carry on further analysis '
'({})'.format(ESSENTIAL_FILES))
parser_gen.add_argument("exp_dir", help="directory containing all the experiment files to be analyzed")
parser_gen.add_argument("-i", "--input_num_records", dest="num_records", type=int,
help="number of input_records to be considered for the generated json context"
"[default: %(default)s]")
parser_gen.add_argument("-c", "--num-cores", dest="num_cores", type=int,
help="number of cores to be considered for the generated json context"
"[default: %(default)s]")
parser_gen.add_argument("-t", "--num-tasks", dest="num_tasks", type=int,
help="number of tasks for each stage"
"[default: %(default)s]")
parser_gen.add_argument("-d", "--deadlines", dest="deadlines", type=int, nargs='+',
help="deadlines to be considered in json context generation"
"[default: %(default)s]")
parser_gen.add_argument("--time-bound", dest="time_bound", type=int, nargs='+',
help="time bounds to be considered in json context generation"
"[default: %(default)s]")
parser_gen.add_argument("-l", "--labeling", dest="labeling", action="store_true", default=False,
help="activates the labeling feature")
parser_gen.add_argument("--d-range", dest="d_range", action="store_true", default=False,
help="Interpret deadline values as a range "
"(it expects 3 values, that will be fed to the range function)")
parser_gen.add_argument("--d-perc", dest="d_percentages", action="store_true", default=False,
help="Generate deadlines that differ by a percentage from the provided value"
"(it expects 1 value)")
parser_gen.add_argument("-v", "--verify", dest="verify", action="store_true",
help="launches verification task of the generated file "
"on a remote server ({})".format(SERVER_NAMES))
parser_gen.add_argument('-s', '--server', choices=SERVER_NAMES,
help='the server where to run verification')
parser_gen.add_argument('-e', '--engine', default='zot',
choices=['zot', 'uppaal'],
help='the verification engine to be used')
parser_gen.add_argument('--plugin', default='ae2sbvzot',
choices=['ae2sbvzot', 'ae2zot'],
help='zot plugin to be used')
parser_gen.add_argument("-w", "--workers", dest="max_workers", type=int, default=MAX_WORKERS,
help="maximum number of verification tasks to be launched"
"[default: %(default)s]")
parser_gen.add_argument('-o', '--search-order',
choices=SEARCH_ORDERS,
help='search order')
parser_ver.add_argument("-j", "--json", help="JSON file to be used for direct verification")
parser_ver.add_argument("-c", "--num-cores", dest="num_cores", type=int,
help="number of cores to be considered for the generated json context"
"[default: %(default)s]")
parser_ver.add_argument("-t", "--num-tasks", dest="num_tasks", type=int,
help="number of tasks for each stage"
"[default: %(default)s]")
parser_ver.add_argument("-d", "--deadlines", dest="deadlines", type=int, nargs='+',
help="deadlines to be considered in json context generation"
"[default: %(default)s]")
parser_ver.add_argument("--time-bound", dest="time_bound", type=int, nargs='+',
help="time bounds to be considered in json context generation"
"[default: %(default)s]")
parser_ver.add_argument("-l", "--labeling", dest="labeling", action="store_true", default=False,
help="activates the labeling feature")
parser_ver.add_argument("-v", "--verify", dest="verify", action="store_true",
help="launches verification task of the generated file "
"on a remote server ({})".format(SERVER_NAMES))
parser_ver.add_argument('-s', '--server', choices=SERVER_NAMES,
help='the server where to run verification')
parser_ver.add_argument('-e', '--engine', default='zot',
choices=['zot', 'uppaal'],
help='the verification engine to be used')
parser_ver.add_argument('--plugin', default='ae2sbvzot',
choices=['ae2sbvzot', 'ae2zot'],
help='zot plugin to be used')
parser_ver.add_argument("-w", "--workers", dest="max_workers", type=int, default=MAX_WORKERS,
help="maximum number of verification tasks to be launched"
"[default: %(default)s]")
parser_ver.add_argument('-o', '--search-order',
choices=SEARCH_ORDERS,
help='search order')
parser_ver.add_argument("--d-range", dest="d_range", action="store_true", default=False,
help="Interpret deadline values as a range "
"(it expects 3 values, that will be fed to the range function)")
parser_ver.add_argument("--d-perc", dest="d_percentages", action="store_true", default=False,
help="Generate deadlines that differ by a percentage from the provided value"
"(it expects 1 value)")
parser_pro.set_defaults(func=pro_runner)
parser_ta.set_defaults(func=time_analysis)
parser_gen.set_defaults(func=generate_spark_context)
parser_ver.set_defaults(func=launch_verification)
args = parser.parse_args()
try:
getattr(args, "func")
except AttributeError:
parser.print_help()
sys.exit(0)
args.func(args)