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
|
#!/usr/bin/env python
# Copyright (C) 2001-2023 Artifex Software, Inc.
# All Rights Reserved.
#
# This software is provided AS-IS with no warranty, either express or
# implied.
#
# This software is distributed under license and may not be copied,
# modified or distributed except as expressly authorized under the terms
# of the license contained in the file LICENSE in this distribution.
#
# Refer to licensing information at http://www.artifex.com or contact
# Artifex Software, Inc., 39 Mesa Street, Suite 108A, San Francisco,
# CA 94129, USA, for further information.
#
# This is a script to parallelize the regression runs "by hand"
# for running on a batch-mode cluster under the PBS queue system.
# it generates a custom testing.cfg and comparefiles directory
# for each node, creates a set of job description files, and runs them.
import os
import string
import re
import sys
import getopt
## globals -- edit these to make it work
run=True # whether to submit the job after creating it
try:
home=os.environ["HOME"]
except KeyError:
home=''
base=os.getcwd()
testdir=home+"/tests/ps/ps3fts" # directory of files to run
configfile="testing.cfg" # template config file
files=os.listdir(testdir) # list of files to run
## defaults -- override from the command line
action='run'
revision=None # revision we're testing, if known
# parse the command line
opts, args = getopt.getopt(sys.argv[1:], "r:", ["rev="])
for o, a in opts:
if o in ("-r", "--rev"):
revision = a
print "parallel run for r" + revision
if len(args):
action = args[0]
## helper functions
def choosecluster():
'''Decide how many nodes of which cluster to run on.
returns a (cluster_name, node_count) tuple.'''
# figure out how many nodes are free
upnodes = os.popen("upnodes")
r = re.compile('^\s+(?P<cluster>\w+).*\s+(?P<procs>\d+)\s+(?P<free>\d+)\s*$')
clusters=[]
nodes = 0
for line in upnodes.readlines():
m = r.match(line)
if m:
name = m.group("cluster")
procs = int(m.group("procs"))
free = int(m.group("free"))
# remember the cluster with the most free nodes
if free > nodes and name != 'total':
nodes = free
cluster = name
clusters.append((name,procs,free))
return (cluster, nodes)
def makepbs(filename):
'''Make a pbs job description file for a command.'''
outfile=open(filename+".pbs","w")
if cluster == 'red' and nodes > 1:
# upnodes reports dual-core nodes twice
outfile.write("#PBS -l nodes=%d:run:%s:ppn=2\n" % (nodes/2,cluster))
else:
outfile.write("#PBS -l nodes=%d:run:%s\n" % (nodes,cluster))
outfile.write("cd %s\n" % base)
outfile.write("mpiexec -comm none ./%s\n" % filename)
outfile.close()
def makepbscleanup(configfiles, comparefiledirs, jobid=None):
'''Make a pbs job to cleanup after another.
Pass in sequences of the files and directories to be removed
and the jobid, if any, that the cleanup should run after.'''
outfile=open("run_regression_cleanup.pbs","w")
outfile.write("#PBS -l nodes=1:run")
# run this on nina by default since it's trivial
#outfile.write(":nina")
if jobid:
outfile.write(" -W depend=afterany:%s\n" % jobid)
else:
outfile.write("\n")
outfile.write("cd %s\n" % base)
for node in range(nodes):
outfile.write("rm -rf " + comparefiledirs[node] + "\n")
outfile.write("rm " + configfiles[node] + "\n")
outfile.close()
def makepbsreport(jobid):
'''Create a report from the output of a previous job
Pass the jobid of the actual regression run so we
know which log to parse.'''
outfile = open("run_regression_report.pbs","w")
outfile.write("#PBS -l nodes=1:run")
# run this on nina by default since it's trivial
#outfile.write(":nina")
# run after the regression job is complete
outfile.write(" -W depend=afterany:%s\n" % jobid)
outfile.write("cd %s\n" % base)
if revision:
dest = " >> regression-r%s.log\n" % revision
else:
dest = " >> regression.%s.log\n" % jobid
# hack: we want the stderr output from the regression run, but
# the jobid pbs appends to the file isn't quite the same as
# what qsub (or qstat) give us.
log = "run_regression.pbs.e%s" % string.split(jobid, '.')[0]
outfile.write("echo Cluster-based regression report BETA -- may not be accurate" + dest)
outfile.write("echo Run split over %d nodes" % nodes + dest)
outfile.write("JOBS=`cat %s | egrep ^Ran | wc -l`" % log + dest)
outfile.write("echo Run completed $JOBS jobs" + dest)
outfile.write("STARTT=`stat -c %Y run_regression.pbs`" + dest)
outfile.write("ENDT=`stat -c %Y " + log + "`" + dest)
outfile.write("echo elapsed time $(($ENDT - $STARTT)) seconds" + dest)
outfile.write("DIFFS=`cat %s | egrep DIFFER$ | wc -l`" % log + dest)
outfile.write("echo Run shows $DIFFS differences" + dest)
outfile.write("echo" + dest)
outfile.write("cat %s | egrep ^Checking | sort" % log + dest)
outfile.write("cat %s | grep 'relevant files'" % log + dest)
outfile.close()
## create a config files from the template for each node
(cluster, nodes) = choosecluster()
print "choosing %s with %d cpus free" % (cluster, nodes)
configfiles=[]
comparefiledirs=[]
print "configuring job..."
for node in range(nodes):
infile=open(configfile)
outfilename=configfile +"."+ str(node)
outfile=open(outfilename,"w")
# remember the filename for later cleanup
configfiles.append(outfilename)
for line in infile.readlines():
try:
key,value=string.split(line)
if key == "comparefiledir":
value=value[:-1]+"."+str(node)+"/"
# remember this for cleanup
comparefiledirs.append(value)
if key == "log_stderr": value=value[:-4]+"."+str(node)+".log"
if key == "log_stdout": value=value[:-4]+"."+str(node)+".log"
outfile.write(key+"\t"+value+"\n")
except ValueError:
outfile.write(line)
nodedir=home+"/comparefiles."+str(node)
# create the per-node directories
os.system("rm -rf " + comparefiledirs[node])
os.mkdir(comparefiledirs[node])
# split the test files into directories for each node
node=0
for file in files:
nodedir=comparefiledirs[node]
os.system("ln -s %s/%s %s/" % (testdir,file,nodedir))
node=node+1
if node >= len(comparefiledirs): node=0
# create our job description files
makepbs("run_regression")
makepbs("make_testdb")
## submit the actual jobs
if run:
# qsub the pbs file
if action == 'update':
job= os.popen("qsub make_testdb.pbs")
else:
job = os.popen("qsub run_regression.pbs")
jobid = string.strip(job.readline())
print "run submitted as", jobid
# append a follow-up report generation job
if action == 'run':
makepbsreport(jobid)
job = os.popen("qsub run_regression_report.pbs")
report_jobid = string.strip(job.readline())
print "report job is", report_jobid
# append a follow-up job to do the cleanup
makepbscleanup(configfiles,comparefiledirs,jobid)
job = os.popen("qsub run_regression_cleanup.pbs")
cleanup_jobid = string.strip(job.readline())
print "cleanup job is", cleanup_jobid
|
