1 |
+ |
""" |
2 |
+ |
Implements the vanilla (local) Condor scheduler |
3 |
+ |
""" |
4 |
+ |
|
5 |
|
__revision__ = "$Id$" |
6 |
|
__version__ = "$Revision$" |
7 |
|
|
8 |
< |
from Scheduler import Scheduler |
9 |
< |
from SchedulerLocal import SchedulerLocal |
6 |
< |
from crab_exceptions import * |
7 |
< |
from crab_util import * |
8 |
< |
from crab_logger import Logger |
9 |
< |
import common |
8 |
> |
from SchedulerLocal import SchedulerLocal |
9 |
> |
from crab_exceptions import CrabException |
10 |
|
|
11 |
+ |
import common |
12 |
|
import os |
13 |
< |
|
14 |
< |
# Naming convention: |
14 |
< |
# methods starting with 'ws' are responsible to provide |
15 |
< |
# corresponding part of the job script ('ws' stands for 'write script'). |
13 |
> |
import socket |
14 |
> |
import sha |
15 |
|
|
16 |
|
class SchedulerCondor(SchedulerLocal) : |
18 |
– |
|
19 |
– |
def __init__(self): |
20 |
– |
Scheduler.__init__(self,"CONDOR") |
21 |
– |
return |
22 |
– |
|
23 |
– |
|
24 |
– |
def configure(self, cfg_params): |
25 |
– |
SchedulerLocal.configure(self, cfg_params) |
26 |
– |
self.environment_unique_identifier ='${HOSTNAME}_${CONDOR_ID}_' + common._db.queryTask('name') |
27 |
– |
|
28 |
– |
try: |
29 |
– |
tmp = cfg_params['CMSSW.datasetpath'] |
30 |
– |
if string.lower(tmp)=='none': |
31 |
– |
self.datasetPath = None |
32 |
– |
self.selectNoInput = 1 |
33 |
– |
else: |
34 |
– |
self.datasetPath = tmp |
35 |
– |
self.selectNoInput = 0 |
36 |
– |
except KeyError: |
37 |
– |
msg = "Error: datasetpath not defined " |
38 |
– |
raise CrabException(msg) |
39 |
– |
|
40 |
– |
return |
41 |
– |
|
42 |
– |
|
43 |
– |
def sched_parameter(self,i,task): |
17 |
|
""" |
18 |
< |
Return scheduler-specific parameters |
19 |
< |
""" |
20 |
< |
index = int(common._db.nJobs()) - 1 |
21 |
< |
sched_param= '' |
22 |
< |
|
23 |
< |
for i in range(index): |
24 |
< |
pass |
25 |
< |
|
26 |
< |
return sched_param |
27 |
< |
|
28 |
< |
|
29 |
< |
def realSchedParams(self,cfg_params): |
30 |
< |
""" |
31 |
< |
Return dictionary with specific parameters, to use |
32 |
< |
with real scheduler |
33 |
< |
""" |
34 |
< |
|
35 |
< |
tmpDir = os.path.join(common.work_space.shareDir(),'.condor_temp') |
36 |
< |
params = {'tmpDir':tmpDir} |
37 |
< |
return params |
38 |
< |
|
39 |
< |
|
40 |
< |
def listMatch(self, seList, full, onlyOSG=True): |
41 |
< |
""" |
42 |
< |
Check the compatibility of available resources |
43 |
< |
""" |
44 |
< |
|
45 |
< |
# May have problems with onlyOSG being false, probably due to lengths of lists and command line. |
46 |
< |
# Either re-write osg_bdii.py with a proper ldap library or break the queries apart |
47 |
< |
|
48 |
< |
if self.selectNoInput: |
49 |
< |
return [True] |
50 |
< |
else: |
51 |
< |
return SchedulerLocal.listMatch(self, seList, full) |
52 |
< |
|
53 |
< |
def decodeLogInfo(self, file): |
54 |
< |
""" |
55 |
< |
Parse logging info file and return main info |
56 |
< |
""" |
57 |
< |
import CondorGLoggingInfo |
58 |
< |
loggingInfo = CondorGLoggingInfo.CondorGLoggingInfo() |
59 |
< |
reason = loggingInfo.decodeReason(file) |
60 |
< |
return reason |
61 |
< |
|
62 |
< |
|
63 |
< |
def wsExitFunc(self): |
64 |
< |
""" |
65 |
< |
""" |
66 |
< |
txt = '\n' |
67 |
< |
txt += '#\n' |
68 |
< |
txt += '# EXECUTE THIS FUNCTION BEFORE EXIT \n' |
69 |
< |
txt += '#\n\n' |
70 |
< |
|
71 |
< |
txt += 'func_exit() { \n' |
72 |
< |
txt += self.wsExitFunc_common() |
73 |
< |
|
74 |
< |
txt += ' tar zcvf ${out_files}.tgz ${final_list}\n' |
75 |
< |
txt += ' cp ${out_files}.tgz $ORIG_WD/\n' |
76 |
< |
txt += ' cp crab_fjr_$NJob.xml $ORIG_WD/\n' |
77 |
< |
|
78 |
< |
txt += ' exit $job_exit_code\n' |
79 |
< |
txt += '}\n' |
80 |
< |
|
81 |
< |
return txt |
82 |
< |
|
83 |
< |
def wsInitialEnvironment(self): |
84 |
< |
""" |
85 |
< |
Returns part of a job script which does scheduler-specific work. |
86 |
< |
""" |
87 |
< |
|
88 |
< |
txt = '\n# Written by SchedulerCondor::wsInitialEnvironment\n' |
89 |
< |
txt += 'echo "Beginning environment"\n' |
90 |
< |
txt += 'printenv | sort\n' |
18 |
> |
Class to implement the vanilla (local) Condor scheduler |
19 |
> |
Naming convention: Methods starting with 'ws' provide |
20 |
> |
the corresponding part of the job script |
21 |
> |
('ws' stands for 'write script'). |
22 |
> |
""" |
23 |
> |
|
24 |
> |
def __init__(self): |
25 |
> |
SchedulerLocal.__init__(self,"CONDOR") |
26 |
> |
self.datasetPath = None |
27 |
> |
self.selectNoInput = None |
28 |
> |
self.environment_unique_identifier = None |
29 |
> |
return |
30 |
> |
|
31 |
> |
|
32 |
> |
def configure(self, cfg_params): |
33 |
> |
""" |
34 |
> |
Configure the scheduler with the config settings from the user |
35 |
> |
""" |
36 |
> |
|
37 |
> |
SchedulerLocal.configure(self, cfg_params) |
38 |
> |
taskHash = sha.new(common._db.queryTask('name')).hexdigest() |
39 |
> |
self.environment_unique_identifier = "https://" + socket.gethostname() + \ |
40 |
> |
'/' + taskHash + "/${NJob}" |
41 |
> |
|
42 |
> |
try: |
43 |
> |
tmp = cfg_params['CMSSW.datasetpath'] |
44 |
> |
if tmp.lower() == 'none': |
45 |
> |
self.datasetPath = None |
46 |
> |
self.selectNoInput = 1 |
47 |
> |
else: |
48 |
> |
self.datasetPath = tmp |
49 |
> |
self.selectNoInput = 0 |
50 |
> |
except KeyError: |
51 |
> |
msg = "Error: datasetpath not defined " |
52 |
> |
raise CrabException(msg) |
53 |
> |
|
54 |
> |
self.return_data = cfg_params.get('USER.return_data', 0) |
55 |
> |
self.copy_data = cfg_params.get("USER.copy_data", 0) |
56 |
> |
self.backup_copy = cfg_params.get('USER.backup_copy',0) |
57 |
> |
|
58 |
> |
if ( int(self.return_data) == 0 and int(self.copy_data) == 0 ): |
59 |
> |
msg = 'Error: return_data and copy_data cannot be set both to 0\n' |
60 |
> |
msg = msg + 'Please modify your crab.cfg file\n' |
61 |
> |
raise CrabException(msg) |
62 |
> |
|
63 |
> |
if ( int(self.return_data) == 1 and int(self.copy_data) == 1 ): |
64 |
> |
msg = 'Error: return_data and copy_data cannot be set both to 1\n' |
65 |
> |
msg = msg + 'Please modify your crab.cfg file\n' |
66 |
> |
raise CrabException(msg) |
67 |
> |
|
68 |
> |
if ( int(self.copy_data) == 0 and int(self.publish_data) == 1 ): |
69 |
> |
msg = 'Warning: publish_data = 1 must be used with copy_data = 1\n' |
70 |
> |
msg = msg + 'Please modify copy_data value in your crab.cfg file\n' |
71 |
> |
common.logger.message(msg) |
72 |
> |
raise CrabException(msg) |
73 |
> |
|
74 |
> |
if ( int(self.copy_data) == 0 and int(self.backup_copy) == 1 ): |
75 |
> |
msg = 'Error: copy_data = 0 and backup_data = 1 ==> to use the backup_copy function, the copy_data value has to be = 1\n' |
76 |
> |
msg = msg + 'Please modify copy_data value in your crab.cfg file\n' |
77 |
> |
raise CrabException(msg) |
78 |
> |
|
79 |
> |
if int(self.copy_data) == 1: |
80 |
> |
self.SE = cfg_params.get('USER.storage_element', None) |
81 |
> |
if not self.SE: |
82 |
> |
msg = "Error. The [USER] section has no 'storage_element'" |
83 |
> |
common.logger.message(msg) |
84 |
> |
raise CrabException(msg) |
85 |
> |
|
86 |
> |
if ( int(self.backup_copy) == 1 and int(self.publish_data) == 1 ): |
87 |
> |
msg = 'Warning: currently the publication is not supported with the backup copy. Work in progress....\n' |
88 |
> |
common.logger.message(msg) |
89 |
> |
raise CrabException(msg) |
90 |
> |
|
91 |
> |
self.proxyValid = 0 |
92 |
> |
self.dontCheckProxy = int(cfg_params.get("EDG.dont_check_proxy",0)) |
93 |
> |
self.proxyServer = cfg_params.get("EDG.proxy_server",'myproxy.cern.ch') |
94 |
> |
common.logger.debug(5,'Setting myproxy server to ' + self.proxyServer) |
95 |
> |
|
96 |
> |
self.group = cfg_params.get("EDG.group", None) |
97 |
> |
self.role = cfg_params.get("EDG.role", None) |
98 |
> |
self.VO = cfg_params.get('EDG.virtual_organization', 'cms') |
99 |
> |
|
100 |
> |
self.checkProxy() |
101 |
> |
self.role = None |
102 |
> |
|
103 |
> |
return |
104 |
> |
|
105 |
> |
|
106 |
> |
def sched_parameter(self, i, task): |
107 |
> |
""" |
108 |
> |
Return scheduler-specific parameters |
109 |
> |
""" |
110 |
> |
|
111 |
> |
index = int(common._db.nJobs()) - 1 |
112 |
> |
schedParam = '' |
113 |
> |
|
114 |
> |
for i in range(index): |
115 |
> |
pass |
116 |
> |
|
117 |
> |
return schedParam |
118 |
> |
|
119 |
> |
|
120 |
> |
def realSchedParams(self, cfg_params): |
121 |
> |
""" |
122 |
> |
Return dictionary with specific parameters, to use with real scheduler |
123 |
> |
""" |
124 |
> |
|
125 |
> |
tmpDir = os.path.join(common.work_space.shareDir(),'.condor_temp') |
126 |
> |
tmpDir = os.path.join(common.work_space.shareDir(),'.condor_temp') |
127 |
> |
jobDir = common.work_space.jobDir() |
128 |
> |
params = {'tmpDir':tmpDir, |
129 |
> |
'jobDir':jobDir} |
130 |
> |
return params |
131 |
> |
|
132 |
> |
|
133 |
> |
def listMatch(self, seList, full): |
134 |
> |
""" |
135 |
> |
Check the compatibility of available resources |
136 |
> |
""" |
137 |
> |
|
138 |
> |
return [True] |
139 |
> |
|
140 |
> |
|
141 |
> |
def decodeLogInfo(self, fileName): |
142 |
> |
""" |
143 |
> |
Parse logging info file and return main info |
144 |
> |
""" |
145 |
> |
|
146 |
> |
import CondorGLoggingInfo |
147 |
> |
loggingInfo = CondorGLoggingInfo.CondorGLoggingInfo() |
148 |
> |
reason = loggingInfo.decodeReason(fileName) |
149 |
> |
return reason |
150 |
> |
|
151 |
> |
|
152 |
> |
def wsCopyOutput(self): |
153 |
> |
""" |
154 |
> |
Write a CopyResults part of a job script, e.g. |
155 |
> |
to copy produced output into a storage element. |
156 |
> |
""" |
157 |
> |
txt = self.wsCopyOutput_comm() |
158 |
> |
return txt |
159 |
> |
|
160 |
> |
|
161 |
> |
def wsExitFunc(self): |
162 |
> |
""" |
163 |
> |
Returns the part of the job script which runs prior to exit |
164 |
> |
""" |
165 |
> |
|
166 |
> |
txt = '\n' |
167 |
> |
txt += '#\n' |
168 |
> |
txt += '# EXECUTE THIS FUNCTION BEFORE EXIT \n' |
169 |
> |
txt += '#\n\n' |
170 |
> |
|
171 |
> |
txt += 'func_exit() { \n' |
172 |
> |
txt += self.wsExitFunc_common() |
173 |
> |
|
174 |
> |
txt += ' tar zcvf ${out_files}.tgz ${final_list}\n' |
175 |
> |
txt += ' cp ${out_files}.tgz $_CONDOR_SCRATCH_DIR/\n' |
176 |
> |
txt += ' cp crab_fjr_$NJob.xml $_CONDOR_SCRATCH_DIR/\n' |
177 |
> |
|
178 |
> |
txt += ' exit $job_exit_code\n' |
179 |
> |
txt += '}\n' |
180 |
> |
|
181 |
> |
return txt |
182 |
> |
|
183 |
> |
def wsInitialEnvironment(self): |
184 |
> |
""" |
185 |
> |
Returns part of a job script which does scheduler-specific work. |
186 |
> |
""" |
187 |
> |
|
188 |
> |
txt = '\n# Written by SchedulerCondor::wsInitialEnvironment\n' |
189 |
> |
txt += 'echo "Beginning environment"\n' |
190 |
> |
txt += 'printenv | sort\n' |
191 |
> |
|
192 |
> |
txt += 'middleware='+self.name()+' \n' |
193 |
> |
txt += 'if [ -e /opt/d-cache/srm/bin ]; then\n' |
194 |
> |
txt += ' export PATH=${PATH}:/opt/d-cache/srm/bin\n' |
195 |
> |
txt += 'fi\n' |
196 |
|
|
197 |
< |
txt += 'middleware='+self.name()+' \n' |
120 |
< |
txt += """ |
197 |
> |
txt += """ |
198 |
|
if [ $_CONDOR_SCRATCH_DIR ] && [ -d $_CONDOR_SCRATCH_DIR ]; then |
199 |
< |
ORIG_WD=`pwd` |
123 |
< |
echo "Change from $ORIG_WD to Condor scratch directory: $_CONDOR_SCRATCH_DIR" |
199 |
> |
echo "cd to Condor scratch directory: $_CONDOR_SCRATCH_DIR" |
200 |
|
if [ -e ../default.tgz ] ;then |
201 |
|
echo "Found ISB in parent directory (Local Condor)" |
202 |
|
cp ../default.tgz $_CONDOR_SCRATCH_DIR |
205 |
|
fi |
206 |
|
""" |
207 |
|
|
208 |
< |
return txt |
208 |
> |
return txt |