dxt_analyzer.py 12.7 KB
Newer Older
Alex Sim's avatar
Alex Sim committed
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
#!/usr/bin/env python2

###
#
# *** Copyright Notice ***
#
# dxt_analyzer.py 
# Copyright (c) 2017, The Regents of the University of California, 
# through Lawrence Berkeley National Laboratory (subject to receipt
# of any required approvals from the U.S. Dept. of Energy).  
# All rights reserved.
#
# If you have questions about your rights to use or distribute this software, 
# please contact Berkeley Lab's Innovation & Partnerships Office 
# at IPO@lbl.gov.
#
# NOTICE.  This software was developed under funding from the 
# U.S. Department of Energy.  As such, the U.S. Government has been granted 
# for itself and others acting on its behalf a paid-up, nonexclusive, 
# irrevocable, worldwide license in the Software to reproduce, prepare 
# derivative works, and perform publicly and display publicly.  Beginning 
# five (5) years after the date permission to assert copyright is obtained 
# from the U.S. Department of Energy, and subject to any subsequent five (5) 
# year renewals, the U.S. Government is granted for itself and others acting 
# on its behalf a paid-up, nonexclusive, irrevocable, worldwide license in 
# the Software to reproduce, prepare derivative works, distribute copies to 
# the public, perform publicly and display publicly, and to permit others to 
# do so.
#
# Email questions to SDMSUPPORT@LBL.GOV
# Scientific Data Management Research Group
# Lawrence Berkeley National Laboratory
#
# last update on Tue Aug  8 08:16:49 PDT 2017
#


'''
dxt_analyzer.py
To plot the read or write activity from Darshan Extended Trace (DXT) logs.

For more information on creating DXT logs, see:
http://www.mcs.anl.gov/research/projects/darshan/docs/darshan3-util.html#_darshan_dxt_parser 

% ./io_activitity_dxt.py --help
usage: dxt_analyzer.py [-h] -i DXT_LOGNAME [-o SAVEFIG] [--show] [--read]
                            [--filemode] [-f FNAME]

io activity plot from dxt log

optional arguments:
  -h, --help            show this help message and exit
  -i DXT_LOGNAME, --input DXT_LOGNAME
                        dxt log path
  -o SAVEFIG, --save SAVEFIG
                        output file name for the plot
  --show                Show the plot rather than saving to a PDF
  --read                READ I/O action to be plotted. 
                        Default is False for WRITE mode.
  --filemode            Single file mode (must be used with --fname)
                        Default is False for all files
  -f FNAME, --fname FNAME
                        name of file to be plotted (must use with --filemode)

e.g.
python dxt_analyzer.py -i darshan_dxt-a.txt
python dxt_analyzer.py -i darshan_dxt-a.txt \
        --filemode -f /global/cscratch1/sd/asim/amrex/a24/plt00000.hdf5
python dxt_analyzer.py -i darshan_dxt-c.txt
python dxt_analyzer.py -i darshan_dxt-d.txt -o dxt-d.pdf
python dxt_analyzer.py -i darshan_dxt-df.txt
python dxt_analyzer.py -i darshan_dxt-v.txt

'''

import numpy as np
import matplotlib
matplotlib.use('PDF')
import matplotlib.pyplot as plt
import re
import argparse

#------------------------------------------------------------------------------
# Regular expression and helper funtion definitions
#------------------------------------------------------------------------------
global_finfo = {} 
#'filename':{'rw':[], 'mount':'', 'fs':'', 'stripe_size':-1, 'stripe_width':-1, 'OSTlist':[]}
#                    Module          rank         write/read      segment               offset                   length                 start                     end                    OST
POSIX_LOG_PATTERN = ' (X_POSIX)\s+([+-]?\d+(?:\.\d+)?)\s+(\S+)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)\s+\[\s*([+-]?\d+(?:\.\d+)?)\]\.*'
POSIX_LOG_NO_OSTS = ' (X_POSIX)\s+([+-]?\d+(?:\.\d+)?)\s+(\S+)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)'

#                    Module          rank         write/read      segment               length                   start                     end 
MPIIO_LOG_PATTERN = ' (X_MPIIO)\s+([+-]?\d+(?:\.\d+)?)\s+(\S+)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)\s+([+-]?\d+(?:\.\d+)?)'

HEADER_PATTERN = '(#\s+)(\S+):(\s+)(\d+)'
COMMENT_PATTERN_DXT = '# \.*'
USEFUL_COMMENT_PATTERN = '# DXT,\.*'
LISTNUM_PATTERN = '(\d+.*)'

header = re.compile(HEADER_PATTERN)
validposix = re.compile(POSIX_LOG_PATTERN)
posix_no_ost = re.compile(POSIX_LOG_NO_OSTS)
validmpiio = re.compile(MPIIO_LOG_PATTERN)
comment_dxt = re.compile(COMMENT_PATTERN_DXT)
use_comment = re.compile(USEFUL_COMMENT_PATTERN)
listnumpattern = re.compile(LISTNUM_PATTERN)


def store_useful_comment(pieces, finfo_dict, curr_fname):
    '''updates a file's dictionary entry using the pieces of a comment/header line from a DXT log'''
    attribute = pieces[0].replace('# DXT, ', '')
    if attribute=='file_id':
        curr_fname = pieces[2].replace(' ','').rstrip('\n')
        if curr_fname not in finfo_dict.keys():
            finfo_dict[curr_fname] = {'mount':'', 'fs':'', 'stripe_size':-1, 'stripe_width':-1, 'OST_list':[]}
        return curr_fname
    elif attribute=='mnt_pt':
        finfo_dict[curr_fname]['mount'] = pieces[1].replace(', fs_type', '').replace(' ', '').rstrip('\n')
        finfo_dict[curr_fname]['fs'] = pieces[2].replace(' ', '').rstrip('\n')
    elif attribute=='Lustre stripe_size':
        finfo_dict[curr_fname]['stripe_size'] = int(pieces[1].replace(', Lustre stripe_count', '').replace(' ', ''))
        finfo_dict[curr_fname]['stripe_width'] = int(pieces[2].replace(' ',''))
    elif attribute=='Lustre OST obdidx':
        liststr = pieces[1][1:]
        listnums = listnumpattern.match(liststr).groups()[0]
        OSTlist = [int(x) for x in re.findall(r'\d+', listnums)]
        finfo_dict[curr_fname]['OST_list'] = OSTlist
    return curr_fname


def parse_dxt_log_line(line, curr_fname, finfo_dict):
    '''takes a DXT log line and stores pertinent file metadata into a dictionary, and parses read/write trace data into a tuple'''
    if not line: return (curr_fname, (),  -1)
    pieces = line.split(":")
    if comment_dxt.match(pieces[0]):
        if use_comment.match(pieces[0]):
            curr_fname = store_useful_comment(pieces, finfo_dict, curr_fname)
            return (curr_fname, (), -1)
        header_match = header.match(line)
        if header_match:
            return (curr_fname, (header_match.group(2),int(header_match.group(4))), -11)
        else:
            return (curr_fname, (), -1)
    data = validposix.match(line)
    if data:
        return (curr_fname,
            (
            data.group(1).replace('X_', '')+'_'+(data.group(3).upper()), 
            (
                int(data.group(2)), #rank
                int(data.group(6)), #length
                float(data.group(7)), #start
                float(data.group(8)), #end
                int(data.group(9))  #OST
            )
        ), 1)
    data = posix_no_ost.match(line)
    if data:
        return (curr_fname,
            (
            data.group(1).replace('X_', '')+'_'+(data.group(3).upper()), 
            (
                int(data.group(2)), #rank
                int(data.group(6)), #length
                float(data.group(7)), #start
                float(data.group(8)), #end
                -1 # No OST
            )
        ), 1)
    data = validmpiio.match(line)
    if data:
        return (curr_fname,
            (
            data.group(1).replace('X_', '')+'_'+(data.group(3).upper()), 
            (
                int(data.group(2)), #rank
                int(data.group(5)), #length
                float(data.group(6)), #start
                float(data.group(7)), #end
            )
        ), 1)
    return (curr_fname, (), -1)


def get_verts_file(data, module, fname, action):
    '''make a list of rectangle vertices to plot the posix/mpi read/write activity for each rank for a specific file'''
    keyword = module+'_'+action
    filedata = map(lambda x: x[1], filter(lambda x: x[0]==fname, data))
    filtered = map(lambda x: x[1], filter(lambda x: x[0]==keyword, filedata))
    activities = map(lambda x: (x[0], (x[2], x[3])), filtered)
    verts = np.zeros((1,4,2))
    for entry in activities:
        lx,rx,by,ty = entry[1][0], entry[1][1], entry[0] - 0.5 , entry[0] + 0.5
        newverts = np.array([((lx,by), (lx,ty), (rx, ty), (rx, by))])
        verts = np.concatenate((verts, newverts))
    return verts[1:]


def get_verts_all(data, module, action):
    '''make a list of rectangle vertices to plot the posix/mpi read/write activity for each rank'''
    keyword = module+'_'+action
    IOdata = map(lambda x: x[1], data)
    filtered = map(lambda x: x[1], filter(lambda x: x[0]==keyword, IOdata))
    activities = map(lambda x: (x[0], (x[2], x[3])), filtered)
    verts = np.zeros((1,4,2))
    for entry in activities:
        lx,rx,by,ty = entry[1][0], entry[1][1], entry[0] - 0.5 , entry[0] + 0.5
        newverts = np.array([((lx,by), (lx,ty), (rx, ty), (rx, by))])
        verts = np.concatenate((verts, newverts))
    return verts[1:]


#------------------------------------------------------------------------------
# Set variables and create plots
#------------------------------------------------------------------------------

# Variables to be set by user
# name of DXT log file
dxt_logname = './darshan_dxt-5967365.txt' 
# create plot for a specific file (mode='file') or for all files (mode='all')
singlefile_mode = False # for 'all' 
mode = 'all' 
# name of file to be plotted (must use with mode='file')
fname = '/global/cscratch1/sd/junmin/heatTxf.n64.s32/ph5.4096.n128s64.t1.h5'
# FLAG for READ I/O action to be plotted.
read_flag = False  # for READ action 
action = 'WRITE' 
# filename to save the plot (must end with .pdf).
# showflag takes priority than savefig.
showflag = False # If you do not want to save, set this to True. 
savefig = 'dxt_plot.pdf' 

parser = argparse.ArgumentParser(description='io activity plot from dxt log')
parser.add_argument("-i", "--input", action="store", dest="dxt_logname", required=True, help="dxt log path")
parser.add_argument("-o", "--save", action="store", dest="savefig", required=False, help="output file name for the plot")
parser.add_argument("--show", action="store_true", dest="showflag", required=False, help="Show the plot rather than saving to a PDF")
parser.add_argument("--read", action="store_true", dest="read_flag", required=False, help="READ I/O action to be plotted. Default is False for WRITE mode.")
parser.add_argument("--filemode", action="store_true", dest="singlefile_mode", required=False, help="Single file mode (must be used with --fname). Default is False for all files")
parser.add_argument("-f", "--fname", action="store", dest="fname", required=False, help="name of file to be plotted (must use with --filemode)")

#args = parser.parse_args(['-l', '/Users/asim/Desktop/simcodes/vpic/runs/ttest/junmin-darshan_dxt-5967365.txt', 
#'--save', 'dxt_plot.pdf'])
args = parser.parse_args()     # uncomment this line for general use

dxt_logname = args.dxt_logname
if (args.savefig):
    savefig = args.savefig
if (args.showflag):
    showflag = True
if (args.read_flag): 
    read_flag = True
    action = 'READ'
if (args.singlefile_mode): 
    singlefile_mode = True
    mode = 'file'
    fname = args.fname
    

with open(dxt_logname) as infile:
    finfo_dict = {}
    curr_fname = ''
    logdata = []
    for line in infile:
        curr_fname, data, flag = parse_dxt_log_line(line, curr_fname, finfo_dict)
        if flag == -11:
            (k,v) = data
            if k == 'jobid': jobid = v
            elif k == 'start_time': start_time = v
            elif k == 'end_time': end_time = v
            elif k == 'nprocs': nprocs = v
        elif flag == 1:
            logdata.append((curr_fname, data))

fig, ax = plt.subplots(dpi=150)

if mode=='file':
    mpiio_verts = get_verts_file(logdata, 'MPIIO', fname, action)
    mpiio_collec = matplotlib.collections.PolyCollection(mpiio_verts, facecolor='blue', edgecolor='blue')
    posix_verts = get_verts_file(logdata, 'POSIX', fname, action)
    posix_collec = matplotlib.collections.PolyCollection(posix_verts, facecolor='red', edgecolor='red')
    title = str(jobid)+'_'+fname.split('/').pop()+'_'+action+'_activity'
else :  # mode=='all'
    mpiio_verts = get_verts_all(logdata, 'MPIIO', action)
    mpiio_collec = matplotlib.collections.PolyCollection(mpiio_verts, facecolor='blue', edgecolor='blue')
    posix_verts = get_verts_all(logdata, 'POSIX', action)
    posix_collec = matplotlib.collections.PolyCollection(posix_verts, facecolor='red', edgecolor='red')
    title = str(jobid)+'_'+action+'_activity'


ax.add_collection(mpiio_collec)
ax.add_collection(posix_collec)
ax.autoscale()
plt.ylabel("MPI rank")
plt.xlabel("Time (s)")
plt.title(title)
if (showflag):
    plt.show()
else:
    plt.savefig(savefig, format='pdf')