1 | #!/usr/bin/env python |
---|
2 | # |
---|
3 | # This file is part of Jobmonarch |
---|
4 | # |
---|
5 | # Copyright (C) 2006-2013 Ramon Bastiaans |
---|
6 | # Copyright (C) 2007, 2009 Dave Love (SGE code) |
---|
7 | # |
---|
8 | # Jobmonarch is free software; you can redistribute it and/or modify |
---|
9 | # it under the terms of the GNU General Public License as published by |
---|
10 | # the Free Software Foundation; either version 2 of the License, or |
---|
11 | # (at your option) any later version. |
---|
12 | # |
---|
13 | # Jobmonarch is distributed in the hope that it will be useful, |
---|
14 | # but WITHOUT ANY WARRANTY; without even the implied warranty of |
---|
15 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
---|
16 | # GNU General Public License for more details. |
---|
17 | # |
---|
18 | # You should have received a copy of the GNU General Public License |
---|
19 | # along with this program; if not, write to the Free Software |
---|
20 | # Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA |
---|
21 | # |
---|
22 | # SVN $Id: jobmond.py 947 2014-01-20 13:23:53Z ramonb $ |
---|
23 | # |
---|
24 | |
---|
25 | # vi :set ts=4 |
---|
26 | |
---|
27 | import sys, getopt, ConfigParser, time, os, socket, string, re |
---|
28 | import xdrlib, socket, syslog, xml, xml.sax, shlex, os.path, pwd |
---|
29 | from xml.sax.handler import feature_namespaces |
---|
30 | from collections import deque |
---|
31 | from glob import glob |
---|
32 | |
---|
33 | VERSION='__VERSION__' |
---|
34 | |
---|
35 | def usage( ver ): |
---|
36 | |
---|
37 | print 'jobmond %s' %VERSION |
---|
38 | |
---|
39 | if ver: |
---|
40 | return 0 |
---|
41 | |
---|
42 | print |
---|
43 | print 'Purpose:' |
---|
44 | print ' The Job Monitoring Daemon (jobmond) reports batch jobs information and statistics' |
---|
45 | print ' to Ganglia, which can be viewed with Job Monarch web frontend' |
---|
46 | print |
---|
47 | print 'Usage: jobmond [OPTIONS]' |
---|
48 | print |
---|
49 | print ' -c, --config=FILE The configuration file to use (default: /etc/jobmond.conf)' |
---|
50 | print ' -p, --pidfile=FILE Use pid file to store the process id' |
---|
51 | print ' -h, --help Print help and exit' |
---|
52 | print ' -v, --version Print version and exit' |
---|
53 | print |
---|
54 | |
---|
55 | def processArgs( args ): |
---|
56 | |
---|
57 | SHORT_L = 'p:hvc:' |
---|
58 | LONG_L = [ 'help', 'config=', 'pidfile=', 'version' ] |
---|
59 | |
---|
60 | global PIDFILE, JOBMOND_CONF |
---|
61 | PIDFILE = None |
---|
62 | |
---|
63 | JOBMOND_CONF = '/etc/jobmond.conf' |
---|
64 | |
---|
65 | try: |
---|
66 | |
---|
67 | opts, args = getopt.getopt( args, SHORT_L, LONG_L ) |
---|
68 | |
---|
69 | except getopt.GetoptError, detail: |
---|
70 | |
---|
71 | print detail |
---|
72 | usage( False ) |
---|
73 | sys.exit( 1 ) |
---|
74 | |
---|
75 | for opt, value in opts: |
---|
76 | |
---|
77 | if opt in [ '--config', '-c' ]: |
---|
78 | |
---|
79 | JOBMOND_CONF = value |
---|
80 | |
---|
81 | if opt in [ '--pidfile', '-p' ]: |
---|
82 | |
---|
83 | PIDFILE = value |
---|
84 | |
---|
85 | if opt in [ '--help', '-h' ]: |
---|
86 | |
---|
87 | usage( False ) |
---|
88 | sys.exit( 0 ) |
---|
89 | |
---|
90 | if opt in [ '--version', '-v' ]: |
---|
91 | |
---|
92 | usage( True ) |
---|
93 | sys.exit( 0 ) |
---|
94 | |
---|
95 | return loadConfig( JOBMOND_CONF ) |
---|
96 | |
---|
97 | html_escape_table = { |
---|
98 | "&": "_", |
---|
99 | '"': "_", |
---|
100 | "'": "_", |
---|
101 | ">": "_", |
---|
102 | "<": "_", |
---|
103 | } |
---|
104 | |
---|
105 | # If this is too slow: |
---|
106 | # - re.sub(r'[\&|\>\<\'\"]','_', variable ) |
---|
107 | # - re.sub(r'[\&\>\<\'\"]','_', variable ) |
---|
108 | # - and even faster: compile regex |
---|
109 | |
---|
110 | def html_escape(text): |
---|
111 | """Produce entities within text.""" |
---|
112 | return string.join((html_escape_table.get(c,c) for c in text), '') |
---|
113 | |
---|
114 | class GangliaConfigParser: |
---|
115 | |
---|
116 | def __init__( self, filename ): |
---|
117 | |
---|
118 | self.conf_lijst = [ ] |
---|
119 | self.conf_dict = { } |
---|
120 | self.filename = filename |
---|
121 | self.file_pointer = file( filename, 'r' ) |
---|
122 | self.lexx = shlex.shlex( self.file_pointer ) |
---|
123 | self.lexx.whitespace_split = True |
---|
124 | |
---|
125 | self.parse() |
---|
126 | |
---|
127 | def __del__( self ): |
---|
128 | |
---|
129 | """ |
---|
130 | Cleanup: close file descriptor |
---|
131 | """ |
---|
132 | |
---|
133 | self.file_pointer.close() |
---|
134 | del self.lexx |
---|
135 | del self.conf_lijst |
---|
136 | |
---|
137 | def removeQuotes( self, value ): |
---|
138 | |
---|
139 | clean_value = value |
---|
140 | clean_value = clean_value.replace( "'", "" ) |
---|
141 | clean_value = clean_value.replace( '"', '' ) |
---|
142 | clean_value = clean_value.strip() |
---|
143 | |
---|
144 | return clean_value |
---|
145 | |
---|
146 | def removeBraces( self, value ): |
---|
147 | |
---|
148 | clean_value = value |
---|
149 | clean_value = clean_value.replace( "(", "" ) |
---|
150 | clean_value = clean_value.replace( ')', '' ) |
---|
151 | clean_value = clean_value.strip() |
---|
152 | |
---|
153 | return clean_value |
---|
154 | |
---|
155 | def parse( self ): |
---|
156 | |
---|
157 | """ |
---|
158 | Parse self.filename using shlex scanning. |
---|
159 | - Removes /* comments */ |
---|
160 | - Traverses (recursively) through all include () statements |
---|
161 | - Stores complete valid config tokens in self.conf_list |
---|
162 | |
---|
163 | i.e.: |
---|
164 | ['globals', |
---|
165 | '{', |
---|
166 | 'daemonize', |
---|
167 | '=', |
---|
168 | 'yes', |
---|
169 | 'setuid', |
---|
170 | '=', |
---|
171 | 'yes', |
---|
172 | 'user', |
---|
173 | '=', |
---|
174 | 'ganglia', |
---|
175 | 'debug_level', |
---|
176 | '=', |
---|
177 | '0', |
---|
178 | <etc> ] |
---|
179 | """ |
---|
180 | |
---|
181 | t = 'bogus' |
---|
182 | c = False |
---|
183 | i = False |
---|
184 | |
---|
185 | while t != self.lexx.eof: |
---|
186 | #print 'get token' |
---|
187 | t = self.lexx.get_token() |
---|
188 | |
---|
189 | if len( t ) >= 2: |
---|
190 | |
---|
191 | if len( t ) >= 4: |
---|
192 | |
---|
193 | if t[:2] == '/*' and t[-2:] == '*/': |
---|
194 | |
---|
195 | #print 'comment line' |
---|
196 | #print 'skipping: %s' %t |
---|
197 | continue |
---|
198 | |
---|
199 | if t == '/*' or t[:2] == '/*': |
---|
200 | c = True |
---|
201 | #print 'comment start' |
---|
202 | #print 'skipping: %s' %t |
---|
203 | continue |
---|
204 | |
---|
205 | if t == '*/' or t[-2:] == '*/': |
---|
206 | c = False |
---|
207 | #print 'skipping: %s' %t |
---|
208 | #print 'comment end' |
---|
209 | continue |
---|
210 | |
---|
211 | if c: |
---|
212 | #print 'skipping: %s' %t |
---|
213 | continue |
---|
214 | |
---|
215 | if t == 'include': |
---|
216 | i = True |
---|
217 | #print 'include start' |
---|
218 | #print 'skipping: %s' %t |
---|
219 | continue |
---|
220 | |
---|
221 | if i: |
---|
222 | |
---|
223 | #print 'include start: %s' %t |
---|
224 | |
---|
225 | t2 = self.removeQuotes( t ) |
---|
226 | t2 = self.removeBraces( t ) |
---|
227 | |
---|
228 | for in_file in glob( self.removeQuotes(t2) ): |
---|
229 | |
---|
230 | #print 'including file: %s' %in_file |
---|
231 | parse_infile = GangliaConfigParser( in_file ) |
---|
232 | |
---|
233 | self.conf_lijst = self.conf_lijst + parse_infile.getConfLijst() |
---|
234 | |
---|
235 | del parse_infile |
---|
236 | |
---|
237 | i = False |
---|
238 | #print 'include end' |
---|
239 | #print 'skipping: %s' %t |
---|
240 | continue |
---|
241 | |
---|
242 | #print 'keep: %s' %t |
---|
243 | self.conf_lijst.append( self.removeQuotes(t) ) |
---|
244 | |
---|
245 | def getConfLijst( self ): |
---|
246 | |
---|
247 | return self.conf_lijst |
---|
248 | |
---|
249 | def confListToDict( self, parent_list=None ): |
---|
250 | |
---|
251 | """ |
---|
252 | Recursively traverses a conf_list and creates dictionary from it |
---|
253 | """ |
---|
254 | |
---|
255 | new_dict = { } |
---|
256 | count = 0 |
---|
257 | skip = 0 |
---|
258 | |
---|
259 | if not parent_list: |
---|
260 | parent_list = self.conf_lijst |
---|
261 | |
---|
262 | #print 'entering confListToDict(): (parent) list size %s' %len(parent_list) |
---|
263 | |
---|
264 | for n, c in enumerate( parent_list ): |
---|
265 | |
---|
266 | count = count + 1 |
---|
267 | |
---|
268 | #print 'CL: n %d c %s' %(n, c) |
---|
269 | |
---|
270 | if skip > 0: |
---|
271 | |
---|
272 | #print '- skipped' |
---|
273 | skip = skip - 1 |
---|
274 | continue |
---|
275 | |
---|
276 | if (n+1) <= (len( parent_list )-1): |
---|
277 | |
---|
278 | if parent_list[(n+1)] == '{': |
---|
279 | |
---|
280 | if not new_dict.has_key( c ): |
---|
281 | new_dict[ c ] = [ ] |
---|
282 | |
---|
283 | (temp_new_dict, skip) = self.confListToDict( parent_list[(n+2):] ) |
---|
284 | new_dict[ c ].append( temp_new_dict ) |
---|
285 | |
---|
286 | if parent_list[(n+1)] == '=' and (n+2) <= (len( parent_list )-1): |
---|
287 | |
---|
288 | if not new_dict.has_key( c ): |
---|
289 | new_dict[ c ] = [ ] |
---|
290 | |
---|
291 | new_dict[ c ].append( parent_list[ (n+2) ] ) |
---|
292 | |
---|
293 | skip = 2 |
---|
294 | |
---|
295 | if parent_list[n] == '}': |
---|
296 | |
---|
297 | #print 'leaving confListToDict(): new dict = %s' %new_dict |
---|
298 | return (new_dict, count) |
---|
299 | |
---|
300 | def makeConfDict( self ): |
---|
301 | |
---|
302 | """ |
---|
303 | Walks through self.conf_list and creates a dictionary based upon config values |
---|
304 | |
---|
305 | i.e.: |
---|
306 | 'tcp_accept_channel': [{'acl': [{'access': [{'action': ['"allow"'], |
---|
307 | 'ip': ['"127.0.0.1"'], |
---|
308 | 'mask': ['32']}]}], |
---|
309 | 'port': ['8649']}], |
---|
310 | 'udp_recv_channel': [{'port': ['8649']}], |
---|
311 | 'udp_send_channel': [{'host': ['145.101.32.3'], |
---|
312 | 'port': ['8649']}, |
---|
313 | {'host': ['145.101.32.207'], |
---|
314 | 'port': ['8649']}]} |
---|
315 | """ |
---|
316 | |
---|
317 | new_dict = { } |
---|
318 | skip = 0 |
---|
319 | |
---|
320 | #print 'entering makeConfDict()' |
---|
321 | |
---|
322 | for n, c in enumerate( self.conf_lijst ): |
---|
323 | |
---|
324 | #print 'M: n %d c %s' %(n, c) |
---|
325 | |
---|
326 | if skip > 0: |
---|
327 | |
---|
328 | #print '- skipped' |
---|
329 | skip = skip - 1 |
---|
330 | continue |
---|
331 | |
---|
332 | if (n+1) <= (len( self.conf_lijst )-1): |
---|
333 | |
---|
334 | if self.conf_lijst[(n+1)] == '{': |
---|
335 | |
---|
336 | if not new_dict.has_key( c ): |
---|
337 | new_dict[ c ] = [ ] |
---|
338 | |
---|
339 | ( temp_new_dict, skip ) = self.confListToDict( self.conf_lijst[(n+2):] ) |
---|
340 | new_dict[ c ].append( temp_new_dict ) |
---|
341 | |
---|
342 | if self.conf_lijst[(n+1)] == '=' and (n+2) <= (len( self.conf_lijst )-1): |
---|
343 | |
---|
344 | if not new_dict.has_key( c ): |
---|
345 | new_dict[ c ] = [ ] |
---|
346 | |
---|
347 | new_dict[ c ].append( self.conf_lijst[ (n+2) ] ) |
---|
348 | |
---|
349 | skip = 2 |
---|
350 | |
---|
351 | self.conf_dict = new_dict |
---|
352 | #print 'leaving makeConfDict(): conf dict size %d' %len( self.conf_dict ) |
---|
353 | |
---|
354 | def checkConfDict( self ): |
---|
355 | |
---|
356 | if len( self.conf_lijst ) == 0: |
---|
357 | |
---|
358 | raise Exception("Something went wrong generating conf list for %s" %self.file_name ) |
---|
359 | |
---|
360 | if len( self.conf_dict ) == 0: |
---|
361 | |
---|
362 | self.makeConfDict() |
---|
363 | |
---|
364 | def getConfDict( self ): |
---|
365 | |
---|
366 | self.checkConfDict() |
---|
367 | return self.conf_dict |
---|
368 | |
---|
369 | def getUdpSendChannels( self ): |
---|
370 | |
---|
371 | self.checkConfDict() |
---|
372 | |
---|
373 | udp_send_channels = [ ] # IP:PORT |
---|
374 | |
---|
375 | if not self.conf_dict.has_key( 'udp_send_channel' ): |
---|
376 | return None |
---|
377 | |
---|
378 | for u in self.conf_dict[ 'udp_send_channel' ]: |
---|
379 | |
---|
380 | if u.has_key( 'mcast_join' ): |
---|
381 | |
---|
382 | ip = u['mcast_join'][0] |
---|
383 | |
---|
384 | elif u.has_key( 'host' ): |
---|
385 | |
---|
386 | ip = u['host'][0] |
---|
387 | |
---|
388 | port = u['port'][0] |
---|
389 | |
---|
390 | udp_send_channels.append( ( ip, port ) ) |
---|
391 | |
---|
392 | if len( udp_send_channels ) == 0: |
---|
393 | return None |
---|
394 | |
---|
395 | return udp_send_channels |
---|
396 | |
---|
397 | def getSectionLastOption( self, section, option ): |
---|
398 | |
---|
399 | """ |
---|
400 | Get last option set in a config section that could be set multiple times in multiple (include) files. |
---|
401 | |
---|
402 | i.e.: getSectionLastOption( 'globals', 'send_metadata_interval' ) |
---|
403 | """ |
---|
404 | |
---|
405 | self.checkConfDict() |
---|
406 | value = None |
---|
407 | |
---|
408 | if not self.conf_dict.has_key( section ): |
---|
409 | |
---|
410 | return None |
---|
411 | |
---|
412 | # Could be set multiple times in multiple (include) files: get last one set |
---|
413 | for c in self.conf_dict[ section ]: |
---|
414 | |
---|
415 | if c.has_key( option ): |
---|
416 | |
---|
417 | value = c[ option ][0] |
---|
418 | |
---|
419 | return value |
---|
420 | |
---|
421 | def getClusterName( self ): |
---|
422 | |
---|
423 | return self.getSectionLastOption( 'cluster', 'name' ) |
---|
424 | |
---|
425 | def getVal( self, section, option ): |
---|
426 | |
---|
427 | return self.getSectionLastOption( section, option ) |
---|
428 | |
---|
429 | def getInt( self, section, valname ): |
---|
430 | |
---|
431 | value = self.getVal( section, valname ) |
---|
432 | |
---|
433 | if not value: |
---|
434 | return None |
---|
435 | |
---|
436 | return int( value ) |
---|
437 | |
---|
438 | def getStr( self, section, valname ): |
---|
439 | |
---|
440 | value = self.getVal( section, valname ) |
---|
441 | |
---|
442 | if not value: |
---|
443 | return None |
---|
444 | |
---|
445 | return str( value ) |
---|
446 | |
---|
447 | def findGmetric(): |
---|
448 | |
---|
449 | for dir in os.path.expandvars( '$PATH' ).split( ':' ): |
---|
450 | |
---|
451 | guess = '%s/%s' %( dir, 'gmetric' ) |
---|
452 | |
---|
453 | if os.path.exists( guess ): |
---|
454 | |
---|
455 | return guess |
---|
456 | |
---|
457 | return False |
---|
458 | |
---|
459 | def loadConfig( filename ): |
---|
460 | |
---|
461 | def getlist( cfg_string ): |
---|
462 | |
---|
463 | my_list = [ ] |
---|
464 | |
---|
465 | for item_txt in cfg_string.split( ',' ): |
---|
466 | |
---|
467 | sep_char = None |
---|
468 | |
---|
469 | item_txt = item_txt.strip() |
---|
470 | |
---|
471 | for s_char in [ "'", '"' ]: |
---|
472 | |
---|
473 | if item_txt.find( s_char ) != -1: |
---|
474 | |
---|
475 | if item_txt.count( s_char ) != 2: |
---|
476 | |
---|
477 | print 'Missing quote: %s' %item_txt |
---|
478 | sys.exit( 1 ) |
---|
479 | |
---|
480 | else: |
---|
481 | |
---|
482 | sep_char = s_char |
---|
483 | break |
---|
484 | |
---|
485 | if sep_char: |
---|
486 | |
---|
487 | item_txt = item_txt.split( sep_char )[1] |
---|
488 | |
---|
489 | my_list.append( item_txt ) |
---|
490 | |
---|
491 | return my_list |
---|
492 | |
---|
493 | if not os.path.isfile( JOBMOND_CONF ): |
---|
494 | |
---|
495 | print "Is not a file or does not exist: '%s'" %JOBMOND_CONF |
---|
496 | sys.exit( 1 ) |
---|
497 | |
---|
498 | try: |
---|
499 | f = open( JOBMOND_CONF, 'r' ) |
---|
500 | except IOError, detail: |
---|
501 | print "Cannot read config file: '%s'" %JOBMOND_CONF |
---|
502 | sys.exit( 1 ) |
---|
503 | else: |
---|
504 | f.close() |
---|
505 | |
---|
506 | cfg = ConfigParser.ConfigParser() |
---|
507 | |
---|
508 | cfg.read( filename ) |
---|
509 | |
---|
510 | global DEBUG_LEVEL, DAEMONIZE, BATCH_SERVER, BATCH_POLL_INTERVAL |
---|
511 | global GMOND_CONF, DETECT_TIME_DIFFS, BATCH_HOST_TRANSLATE |
---|
512 | global BATCH_API, QUEUE, GMETRIC_TARGET, USE_SYSLOG |
---|
513 | global SYSLOG_LEVEL, SYSLOG_FACILITY, GMETRIC_BINARY |
---|
514 | global METRIC_MAX_VAL_LEN, GMOND_UDP_SEND_CHANNELS |
---|
515 | |
---|
516 | DEBUG_LEVEL = cfg.getint( 'DEFAULT', 'DEBUG_LEVEL' ) |
---|
517 | |
---|
518 | DAEMONIZE = cfg.getboolean( 'DEFAULT', 'DAEMONIZE' ) |
---|
519 | |
---|
520 | SYSLOG_LEVEL = -1 |
---|
521 | SYSLOG_FACILITY = None |
---|
522 | |
---|
523 | try: |
---|
524 | USE_SYSLOG = cfg.getboolean( 'DEFAULT', 'USE_SYSLOG' ) |
---|
525 | |
---|
526 | except ConfigParser.NoOptionError: |
---|
527 | |
---|
528 | USE_SYSLOG = True |
---|
529 | |
---|
530 | debug_msg( 0, 'ERROR: no option USE_SYSLOG found: assuming yes' ) |
---|
531 | |
---|
532 | if USE_SYSLOG: |
---|
533 | |
---|
534 | try: |
---|
535 | SYSLOG_LEVEL = cfg.getint( 'DEFAULT', 'SYSLOG_LEVEL' ) |
---|
536 | |
---|
537 | except ConfigParser.NoOptionError: |
---|
538 | |
---|
539 | debug_msg( 0, 'ERROR: no option SYSLOG_LEVEL found: assuming level 0' ) |
---|
540 | SYSLOG_LEVEL = 0 |
---|
541 | |
---|
542 | try: |
---|
543 | |
---|
544 | SYSLOG_FACILITY = eval( 'syslog.LOG_' + cfg.get( 'DEFAULT', 'SYSLOG_FACILITY' ) ) |
---|
545 | |
---|
546 | except ConfigParser.NoOptionError: |
---|
547 | |
---|
548 | SYSLOG_FACILITY = syslog.LOG_DAEMON |
---|
549 | |
---|
550 | debug_msg( 0, 'ERROR: no option SYSLOG_FACILITY found: assuming facility DAEMON' ) |
---|
551 | |
---|
552 | try: |
---|
553 | |
---|
554 | BATCH_SERVER = cfg.get( 'DEFAULT', 'BATCH_SERVER' ) |
---|
555 | |
---|
556 | except ConfigParser.NoOptionError: |
---|
557 | |
---|
558 | # Not required for all API's: only pbs api allows remote connections |
---|
559 | BATCH_SERVER = None |
---|
560 | |
---|
561 | try: |
---|
562 | |
---|
563 | BATCH_POLL_INTERVAL = cfg.getint( 'DEFAULT', 'BATCH_POLL_INTERVAL' ) |
---|
564 | |
---|
565 | except ConfigParser.NoOptionError: |
---|
566 | |
---|
567 | # Backwards compatibility for old configs |
---|
568 | # |
---|
569 | |
---|
570 | BATCH_POLL_INTERVAL = cfg.getint( 'DEFAULT', 'TORQUE_POLL_INTERVAL' ) |
---|
571 | api_guess = 'pbs' |
---|
572 | |
---|
573 | try: |
---|
574 | |
---|
575 | GMOND_CONF = cfg.get( 'DEFAULT', 'GMOND_CONF' ) |
---|
576 | |
---|
577 | except ConfigParser.NoOptionError: |
---|
578 | |
---|
579 | # Not specified: assume /etc/ganglia/gmond.conf |
---|
580 | # |
---|
581 | GMOND_CONF = '/etc/ganglia/gmond.conf' |
---|
582 | |
---|
583 | ganglia_cfg = GangliaConfigParser( GMOND_CONF ) |
---|
584 | GMETRIC_TARGET = None |
---|
585 | |
---|
586 | GMOND_UDP_SEND_CHANNELS = ganglia_cfg.getUdpSendChannels() |
---|
587 | |
---|
588 | if not GMOND_UDP_SEND_CHANNELS: |
---|
589 | |
---|
590 | debug_msg( 0, "WARNING: Can't parse udp_send_channel from: '%s' - Trying: %s" %( GMOND_CONF, JOBMOND_CONF ) ) |
---|
591 | |
---|
592 | # Couldn't figure it out: let's see if it's in our jobmond.conf |
---|
593 | # |
---|
594 | try: |
---|
595 | |
---|
596 | GMETRIC_TARGET = cfg.get( 'DEFAULT', 'GMETRIC_TARGET' ) |
---|
597 | |
---|
598 | # Guess not: now just give up |
---|
599 | |
---|
600 | except ConfigParser.NoOptionError: |
---|
601 | |
---|
602 | GMETRIC_TARGET = None |
---|
603 | |
---|
604 | debug_msg( 0, "ERROR: GMETRIC_TARGET not set: internal Gmetric handling aborted. Failing back to DEPRECATED use of gmond.conf/gmetric binary. This will slow down jobmond significantly!" ) |
---|
605 | |
---|
606 | gmetric_bin = findGmetric() |
---|
607 | |
---|
608 | if gmetric_bin: |
---|
609 | |
---|
610 | GMETRIC_BINARY = gmetric_bin |
---|
611 | else: |
---|
612 | debug_msg( 0, "WARNING: Can't find gmetric binary anywhere in $PATH" ) |
---|
613 | |
---|
614 | try: |
---|
615 | |
---|
616 | GMETRIC_BINARY = cfg.get( 'DEFAULT', 'GMETRIC_BINARY' ) |
---|
617 | |
---|
618 | except ConfigParser.NoOptionError: |
---|
619 | |
---|
620 | print "FATAL ERROR: GMETRIC_BINARY not set and not in $PATH" |
---|
621 | sys.exit( 1 ) |
---|
622 | |
---|
623 | #TODO: is this really still needed or should be automatic |
---|
624 | DETECT_TIME_DIFFS = cfg.getboolean( 'DEFAULT', 'DETECT_TIME_DIFFS' ) |
---|
625 | |
---|
626 | try: |
---|
627 | BATCH_HOST_TRANSLATE = getlist( cfg.get( 'DEFAULT', 'BATCH_HOST_TRANSLATE' ) ) |
---|
628 | |
---|
629 | except ConfigParser.NoOptionError: |
---|
630 | |
---|
631 | BATCH_HOST_TRANSLATE = [ ] |
---|
632 | pass |
---|
633 | |
---|
634 | try: |
---|
635 | |
---|
636 | BATCH_API = cfg.get( 'DEFAULT', 'BATCH_API' ) |
---|
637 | |
---|
638 | except ConfigParser.NoOptionError, detail: |
---|
639 | |
---|
640 | print "FATAL ERROR: BATCH_API not set" |
---|
641 | sys.exit( 1 ) |
---|
642 | |
---|
643 | try: |
---|
644 | |
---|
645 | QUEUE = getlist( cfg.get( 'DEFAULT', 'QUEUE' ) ) |
---|
646 | |
---|
647 | except ConfigParser.NoOptionError, detail: |
---|
648 | |
---|
649 | QUEUE = None |
---|
650 | |
---|
651 | METRIC_MAX_VAL_LEN = ganglia_cfg.getInt( 'globals', 'max_udp_msg_len' ) |
---|
652 | |
---|
653 | return True |
---|
654 | |
---|
655 | def fqdn_parts (fqdn): |
---|
656 | |
---|
657 | """Return pair of host and domain for fully-qualified domain name arg.""" |
---|
658 | |
---|
659 | parts = fqdn.split (".") |
---|
660 | |
---|
661 | return (parts[0], string.join(parts[1:], ".")) |
---|
662 | |
---|
663 | class DataProcessor: |
---|
664 | |
---|
665 | """Class for processing of data""" |
---|
666 | |
---|
667 | binary = None |
---|
668 | |
---|
669 | def __init__( self, binary=None ): |
---|
670 | |
---|
671 | """Remember alternate binary location if supplied""" |
---|
672 | |
---|
673 | global GMETRIC_BINARY, GMOND_CONF |
---|
674 | |
---|
675 | if binary: |
---|
676 | self.binary = binary |
---|
677 | |
---|
678 | if not self.binary and not GMETRIC_TARGET and not GMOND_UDP_SEND_CHANNELS: |
---|
679 | self.binary = GMETRIC_BINARY |
---|
680 | |
---|
681 | # Timeout for XML |
---|
682 | # |
---|
683 | # From ganglia's documentation: |
---|
684 | # |
---|
685 | # 'A metric will be deleted DMAX seconds after it is received, and |
---|
686 | # DMAX=0 means eternal life.' |
---|
687 | |
---|
688 | self.dmax = str( int( int( BATCH_POLL_INTERVAL ) * 2 ) ) |
---|
689 | |
---|
690 | if GMOND_CONF and not GMETRIC_TARGET and not GMOND_UDP_SEND_CHANNELS: |
---|
691 | |
---|
692 | incompatible = self.checkGmetricVersion() |
---|
693 | |
---|
694 | if incompatible: |
---|
695 | |
---|
696 | print 'Ganglia/Gmetric version not compatible, please upgrade to at least 3.3.8' |
---|
697 | sys.exit( 1 ) |
---|
698 | |
---|
699 | self.gmetric_send_instances = { } |
---|
700 | |
---|
701 | |
---|
702 | def __del__( self ): |
---|
703 | |
---|
704 | for send_instance in self.gmetric_send_instances: |
---|
705 | |
---|
706 | del send_instance |
---|
707 | |
---|
708 | def checkGmetricVersion( self ): |
---|
709 | |
---|
710 | """ |
---|
711 | Check version of gmetric is at least 3.3.8 |
---|
712 | for the syntax we use |
---|
713 | """ |
---|
714 | |
---|
715 | global METRIC_MAX_VAL_LEN, GMETRIC_TARGET |
---|
716 | |
---|
717 | incompatible = 0 |
---|
718 | |
---|
719 | gfp = os.popen( self.binary + ' --version' ) |
---|
720 | lines = gfp.readlines() |
---|
721 | |
---|
722 | gfp.close() |
---|
723 | |
---|
724 | for line in lines: |
---|
725 | |
---|
726 | line = line.split( ' ' ) |
---|
727 | |
---|
728 | if len( line ) == 2 and str( line ).find( 'gmetric' ) != -1: |
---|
729 | |
---|
730 | gmetric_version = line[1].split( '\n' )[0] |
---|
731 | |
---|
732 | version_major = int( gmetric_version.split( '.' )[0] ) |
---|
733 | version_minor = int( gmetric_version.split( '.' )[1] ) |
---|
734 | version_patch = int( gmetric_version.split( '.' )[2] ) |
---|
735 | |
---|
736 | incompatible = 0 |
---|
737 | |
---|
738 | if version_major < 3: |
---|
739 | |
---|
740 | incompatible = 1 |
---|
741 | |
---|
742 | elif version_major == 3: |
---|
743 | |
---|
744 | if version_minor < 3: |
---|
745 | |
---|
746 | incompatible = 1 |
---|
747 | |
---|
748 | elif version_patch < 8: |
---|
749 | |
---|
750 | incompatible = 1 |
---|
751 | |
---|
752 | return incompatible |
---|
753 | |
---|
754 | def multicastGmetric( self, metricname, metricval, valtype='string', units='' ): |
---|
755 | |
---|
756 | """Call gmetric binary and multicast""" |
---|
757 | |
---|
758 | cmd = self.binary |
---|
759 | |
---|
760 | if GMOND_UDP_SEND_CHANNELS: |
---|
761 | |
---|
762 | for c_ip, c_port in GMOND_UDP_SEND_CHANNELS: |
---|
763 | |
---|
764 | metric_debug = "[gmetric %s:%s] name: %s - val: %s - dmax: %s" %( str(c_ip), str(c_port), str( metricname ), str( metricval ), str( self.dmax ) ) |
---|
765 | |
---|
766 | debug_msg( 10, printTime() + ' ' + metric_debug) |
---|
767 | |
---|
768 | if not self.gmetric_send_instances.has_key( (c_ip, c_port) ): |
---|
769 | |
---|
770 | self.gmetric_send_instances[ (c_ip, c_port) ] = Gmetric( c_ip, c_port ) |
---|
771 | |
---|
772 | self.gmetric_send_instances[ (c_ip, c_port) ].send( str( metricname ), str( metricval ), str( self.dmax ), valtype, units ) |
---|
773 | |
---|
774 | elif GMETRIC_TARGET: |
---|
775 | |
---|
776 | GMETRIC_TARGET_HOST = GMETRIC_TARGET.split( ':' )[0] |
---|
777 | GMETRIC_TARGET_PORT = GMETRIC_TARGET.split( ':' )[1] |
---|
778 | |
---|
779 | metric_debug = "[gmetric] name: %s - val: %s - dmax: %s" %( str( metricname ), str( metricval ), str( self.dmax ) ) |
---|
780 | |
---|
781 | debug_msg( 10, printTime() + ' ' + metric_debug) |
---|
782 | |
---|
783 | if not self.gmetric_send_instances.has_key( (GMETRIC_TARGET_HOST, GMETRIC_TARGET_PORT) ): |
---|
784 | |
---|
785 | self.gmetric_send_instances[ (GMETRIC_TARGET_HOST, GMETRIC_TARGET_PORT) ] = Gmetric( GMETRIC_TARGET_HOST, GMETRIC_TARGET_PORT ) |
---|
786 | |
---|
787 | self.gmetric_send_instances[ (GMETRIC_TARGET_HOST, GMETRIC_TARGET_PORT) ].send( str( metricname ), str( metricval ), str( self.dmax ), valtype, units ) |
---|
788 | |
---|
789 | else: |
---|
790 | try: |
---|
791 | cmd = cmd + ' -c' + GMOND_CONF |
---|
792 | |
---|
793 | except NameError: |
---|
794 | |
---|
795 | debug_msg( 10, 'Assuming /etc/ganglia/gmond.conf for gmetric cmd' ) |
---|
796 | |
---|
797 | cmd = cmd + ' -n' + str( metricname )+ ' -v"' + str( metricval )+ '" -t' + str( valtype ) + ' -d' + str( self.dmax ) |
---|
798 | |
---|
799 | if len( units ) > 0: |
---|
800 | |
---|
801 | cmd = cmd + ' -u"' + units + '"' |
---|
802 | |
---|
803 | debug_msg( 10, printTime() + ' ' + cmd ) |
---|
804 | |
---|
805 | os.system( cmd ) |
---|
806 | |
---|
807 | class DataGatherer: |
---|
808 | |
---|
809 | """Skeleton class for batch system DataGatherer""" |
---|
810 | |
---|
811 | def printJobs( self, jobs ): |
---|
812 | |
---|
813 | """Print a jobinfo overview""" |
---|
814 | |
---|
815 | for name, attrs in self.jobs.items(): |
---|
816 | |
---|
817 | print 'job %s' %(name) |
---|
818 | |
---|
819 | for name, val in attrs.items(): |
---|
820 | |
---|
821 | print '\t%s = %s' %( name, val ) |
---|
822 | |
---|
823 | def printJob( self, jobs, job_id ): |
---|
824 | |
---|
825 | """Print job with job_id from jobs""" |
---|
826 | |
---|
827 | print 'job %s' %(job_id) |
---|
828 | |
---|
829 | for name, val in jobs[ job_id ].items(): |
---|
830 | |
---|
831 | print '\t%s = %s' %( name, val ) |
---|
832 | |
---|
833 | def getAttr( self, attrs, name ): |
---|
834 | |
---|
835 | """Return certain attribute from dictionary, if exists""" |
---|
836 | |
---|
837 | if attrs.has_key( name ): |
---|
838 | |
---|
839 | return attrs[ name ] |
---|
840 | else: |
---|
841 | return '' |
---|
842 | |
---|
843 | def jobDataChanged( self, jobs, job_id, attrs ): |
---|
844 | |
---|
845 | """Check if job with attrs and job_id in jobs has changed""" |
---|
846 | |
---|
847 | if jobs.has_key( job_id ): |
---|
848 | |
---|
849 | oldData = jobs[ job_id ] |
---|
850 | else: |
---|
851 | return 1 |
---|
852 | |
---|
853 | for name, val in attrs.items(): |
---|
854 | |
---|
855 | if oldData.has_key( name ): |
---|
856 | |
---|
857 | if oldData[ name ] != attrs[ name ]: |
---|
858 | |
---|
859 | return 1 |
---|
860 | |
---|
861 | else: |
---|
862 | return 1 |
---|
863 | |
---|
864 | return 0 |
---|
865 | |
---|
866 | def submitJobData( self ): |
---|
867 | |
---|
868 | """Submit job info list""" |
---|
869 | |
---|
870 | global BATCH_API |
---|
871 | |
---|
872 | self.dp.multicastGmetric( 'zplugin_monarch_heartbeat', str( int( int( self.cur_time ) + int( self.timeoffset ) ) ) ) |
---|
873 | |
---|
874 | running_jobs = 0 |
---|
875 | queued_jobs = 0 |
---|
876 | |
---|
877 | # Count how many running/queued jobs we found |
---|
878 | # |
---|
879 | for jobid, jobattrs in self.jobs.items(): |
---|
880 | |
---|
881 | if jobattrs[ 'status' ] == 'Q': |
---|
882 | |
---|
883 | queued_jobs += 1 |
---|
884 | |
---|
885 | elif jobattrs[ 'status' ] == 'R': |
---|
886 | |
---|
887 | running_jobs += 1 |
---|
888 | |
---|
889 | # Report running/queued jobs as seperate metric for a nice RRD graph |
---|
890 | # |
---|
891 | self.dp.multicastGmetric( 'zplugin_monarch_rj', str( running_jobs ), 'uint32', 'jobs' ) |
---|
892 | self.dp.multicastGmetric( 'zplugin_monarch_qj', str( queued_jobs ), 'uint32', 'jobs' ) |
---|
893 | |
---|
894 | # Report down/offline nodes in batch (PBS only ATM) |
---|
895 | # |
---|
896 | if BATCH_API in [ 'pbs', 'slurm' ]: |
---|
897 | |
---|
898 | domain = fqdn_parts( socket.getfqdn() )[1] |
---|
899 | |
---|
900 | downed_nodes = list() |
---|
901 | offline_nodes = list() |
---|
902 | |
---|
903 | l = ['state'] |
---|
904 | |
---|
905 | nodelist = self.getNodeData() |
---|
906 | |
---|
907 | for name, node in nodelist.items(): |
---|
908 | |
---|
909 | if ( node[ 'state' ].find( "down" ) != -1 ): |
---|
910 | |
---|
911 | downed_nodes.append( name ) |
---|
912 | |
---|
913 | if ( node[ 'state' ].find( "offline" ) != -1 ): |
---|
914 | |
---|
915 | offline_nodes.append( name ) |
---|
916 | |
---|
917 | downnodeslist = do_nodelist( downed_nodes ) |
---|
918 | offlinenodeslist = do_nodelist( offline_nodes ) |
---|
919 | |
---|
920 | down_str = 'nodes=%s domain=%s reported=%s' %( string.join( downnodeslist, ';' ), domain, str( int( int( self.cur_time ) + int( self.timeoffset ) ) ) ) |
---|
921 | offl_str = 'nodes=%s domain=%s reported=%s' %( string.join( offlinenodeslist, ';' ), domain, str( int( int( self.cur_time ) + int( self.timeoffset ) ) ) ) |
---|
922 | self.dp.multicastGmetric( 'zplugin_monarch_down' , down_str ) |
---|
923 | self.dp.multicastGmetric( 'zplugin_monarch_offline', offl_str ) |
---|
924 | |
---|
925 | # Now let's spread the knowledge |
---|
926 | # |
---|
927 | for jobid, jobattrs in self.jobs.items(): |
---|
928 | |
---|
929 | # Make gmetric values for each job: respect max gmetric value length |
---|
930 | # |
---|
931 | gmetric_val = self.compileGmetricVal( jobid, jobattrs ) |
---|
932 | metric_increment = 0 |
---|
933 | |
---|
934 | # If we have more job info than max gmetric value length allows, split it up |
---|
935 | # amongst multiple metrics |
---|
936 | # |
---|
937 | for val in gmetric_val: |
---|
938 | |
---|
939 | metric_name = 'zplugin_monarch_job_%s_%s' %( str(metric_increment) , str( jobid ) ) |
---|
940 | self.dp.multicastGmetric( metric_name, val ) |
---|
941 | |
---|
942 | # Increase follow number if this jobinfo is split up amongst more than 1 gmetric |
---|
943 | # |
---|
944 | metric_increment = metric_increment + 1 |
---|
945 | |
---|
946 | def compileGmetricVal( self, jobid, jobattrs ): |
---|
947 | |
---|
948 | """Create a val string for gmetric of jobinfo""" |
---|
949 | |
---|
950 | gval_lists = [ ] |
---|
951 | val_list = { } |
---|
952 | |
---|
953 | for val_name, val_value in jobattrs.items(): |
---|
954 | |
---|
955 | # These are our own metric names, i.e.: status, start_timestamp, etc |
---|
956 | # |
---|
957 | val_list_names_len = len( string.join( val_list.keys() ) ) + len(val_list.keys()) |
---|
958 | |
---|
959 | # These are their corresponding values |
---|
960 | # |
---|
961 | val_list_vals_len = len( string.join( val_list.values() ) ) + len(val_list.values()) |
---|
962 | |
---|
963 | if val_name == 'nodes' and jobattrs['status'] == 'R': |
---|
964 | |
---|
965 | node_str = None |
---|
966 | |
---|
967 | for node in val_value: |
---|
968 | |
---|
969 | if node_str: |
---|
970 | |
---|
971 | node_str = node_str + ';' + node |
---|
972 | else: |
---|
973 | node_str = node |
---|
974 | |
---|
975 | # Make sure if we add this new info, that the total metric's value length does not exceed METRIC_MAX_VAL_LEN |
---|
976 | # |
---|
977 | if (val_list_names_len + len(val_name) ) + (val_list_vals_len + len(node_str) ) > METRIC_MAX_VAL_LEN: |
---|
978 | |
---|
979 | # It's too big, we need to make a new gmetric for the additional info |
---|
980 | # |
---|
981 | val_list[ val_name ] = node_str |
---|
982 | |
---|
983 | gval_lists.append( val_list ) |
---|
984 | |
---|
985 | val_list = { } |
---|
986 | node_str = None |
---|
987 | |
---|
988 | val_list[ val_name ] = node_str |
---|
989 | |
---|
990 | gval_lists.append( val_list ) |
---|
991 | |
---|
992 | val_list = { } |
---|
993 | |
---|
994 | elif val_value != '': |
---|
995 | |
---|
996 | # Make sure if we add this new info, that the total metric's value length does not exceed METRIC_MAX_VAL_LEN |
---|
997 | # |
---|
998 | if (val_list_names_len + len(val_name) ) + (val_list_vals_len + len(str(val_value)) ) > METRIC_MAX_VAL_LEN: |
---|
999 | |
---|
1000 | # It's too big, we need to make a new gmetric for the additional info |
---|
1001 | # |
---|
1002 | gval_lists.append( val_list ) |
---|
1003 | |
---|
1004 | val_list = { } |
---|
1005 | |
---|
1006 | val_list[ val_name ] = val_value |
---|
1007 | |
---|
1008 | if len( val_list ) > 0: |
---|
1009 | |
---|
1010 | gval_lists.append( val_list ) |
---|
1011 | |
---|
1012 | str_list = [ ] |
---|
1013 | |
---|
1014 | # Now append the value names and values together, i.e.: stop_timestamp=value, etc |
---|
1015 | # |
---|
1016 | for val_list in gval_lists: |
---|
1017 | |
---|
1018 | my_val_str = None |
---|
1019 | |
---|
1020 | for val_name, val_value in val_list.items(): |
---|
1021 | |
---|
1022 | if type(val_value) == list: |
---|
1023 | |
---|
1024 | val_value = val_value.join( ',' ) |
---|
1025 | |
---|
1026 | if my_val_str: |
---|
1027 | |
---|
1028 | try: |
---|
1029 | # fixme: It's getting |
---|
1030 | # ('nodes', None) items |
---|
1031 | my_val_str = my_val_str + ' ' + val_name + '=' + html_escape(val_value) |
---|
1032 | except: |
---|
1033 | pass |
---|
1034 | |
---|
1035 | else: |
---|
1036 | my_val_str = val_name + '=' + html_escape(val_value) |
---|
1037 | |
---|
1038 | str_list.append( my_val_str ) |
---|
1039 | |
---|
1040 | return str_list |
---|
1041 | |
---|
1042 | def daemon( self ): |
---|
1043 | |
---|
1044 | """Run as daemon forever""" |
---|
1045 | |
---|
1046 | # Fork the first child |
---|
1047 | # |
---|
1048 | pid = os.fork() |
---|
1049 | if pid > 0: |
---|
1050 | sys.exit(0) # end parent |
---|
1051 | |
---|
1052 | # creates a session and sets the process group ID |
---|
1053 | # |
---|
1054 | os.setsid() |
---|
1055 | |
---|
1056 | # Fork the second child |
---|
1057 | # |
---|
1058 | pid = os.fork() |
---|
1059 | if pid > 0: |
---|
1060 | sys.exit(0) # end parent |
---|
1061 | |
---|
1062 | write_pidfile() |
---|
1063 | |
---|
1064 | # Go to the root directory and set the umask |
---|
1065 | # |
---|
1066 | os.chdir('/') |
---|
1067 | os.umask(0) |
---|
1068 | |
---|
1069 | sys.stdin.close() |
---|
1070 | sys.stdout.close() |
---|
1071 | sys.stderr.close() |
---|
1072 | |
---|
1073 | os.open('/dev/null', os.O_RDWR) |
---|
1074 | os.dup2(0, 1) |
---|
1075 | os.dup2(0, 2) |
---|
1076 | |
---|
1077 | self.run() |
---|
1078 | |
---|
1079 | def run( self ): |
---|
1080 | |
---|
1081 | """Main thread""" |
---|
1082 | |
---|
1083 | while ( 1 ): |
---|
1084 | |
---|
1085 | self.getJobData() |
---|
1086 | self.submitJobData() |
---|
1087 | time.sleep( BATCH_POLL_INTERVAL ) |
---|
1088 | |
---|
1089 | # SGE code by Dave Love <fx@gnu.org>. Tested with SGE 6.0u8 and 6.0u11. May |
---|
1090 | # work with SGE 6.1 (else should be easily fixable), but definitely doesn't |
---|
1091 | # with 6.2. See also the fixmes. |
---|
1092 | |
---|
1093 | class NoJobs (Exception): |
---|
1094 | """Exception raised by empty job list in qstat output.""" |
---|
1095 | pass |
---|
1096 | |
---|
1097 | class SgeQstatXMLParser(xml.sax.handler.ContentHandler): |
---|
1098 | """SAX handler for XML output from Sun Grid Engine's `qstat'.""" |
---|
1099 | |
---|
1100 | def __init__(self): |
---|
1101 | self.value = "" |
---|
1102 | self.joblist = [] |
---|
1103 | self.job = {} |
---|
1104 | self.queue = "" |
---|
1105 | self.in_joblist = False |
---|
1106 | self.lrequest = False |
---|
1107 | self.eltq = deque() |
---|
1108 | xml.sax.handler.ContentHandler.__init__(self) |
---|
1109 | |
---|
1110 | # The structure of the output is as follows (for SGE 6.0). It's |
---|
1111 | # similar for 6.1, but radically different for 6.2, and is |
---|
1112 | # undocumented generally. Unfortunately it's voluminous, and probably |
---|
1113 | # doesn't scale to large clusters/queues. |
---|
1114 | |
---|
1115 | # <detailed_job_info xmlns:xsd="http://www.w3.org/2001/XMLSchema"> |
---|
1116 | # <djob_info> |
---|
1117 | # <qmaster_response> <!-- job --> |
---|
1118 | # ... |
---|
1119 | # <JB_ja_template> |
---|
1120 | # <ulong_sublist> |
---|
1121 | # ... <!-- start_time, state ... --> |
---|
1122 | # </ulong_sublist> |
---|
1123 | # </JB_ja_template> |
---|
1124 | # <JB_ja_tasks> |
---|
1125 | # <ulong_sublist> |
---|
1126 | # ... <!-- task info |
---|
1127 | # </ulong_sublist> |
---|
1128 | # ... |
---|
1129 | # </JB_ja_tasks> |
---|
1130 | # ... |
---|
1131 | # </qmaster_response> |
---|
1132 | # </djob_info> |
---|
1133 | # <messages> |
---|
1134 | # ... |
---|
1135 | |
---|
1136 | # NB. We might treat each task as a separate job, like |
---|
1137 | # straight qstat output, but the web interface expects jobs to |
---|
1138 | # be identified by integers, not, say, <job number>.<task>. |
---|
1139 | |
---|
1140 | # So, I lied. If the job list is empty, we get invalid XML |
---|
1141 | # like this, which we need to defend against: |
---|
1142 | |
---|
1143 | # <unknown_jobs xmlns:xsd="http://www.w3.org/2001/XMLSchema"> |
---|
1144 | # <> |
---|
1145 | # <ST_name>*</ST_name> |
---|
1146 | # </> |
---|
1147 | # </unknown_jobs> |
---|
1148 | |
---|
1149 | def startElement(self, name, attrs): |
---|
1150 | self.value = "" |
---|
1151 | if name == "djob_info": # job list |
---|
1152 | self.in_joblist = True |
---|
1153 | # The job container is "qmaster_response" in SGE 6.0 |
---|
1154 | # and 6.1, but "element" in 6.2. This is only the very |
---|
1155 | # start of what's necessary for 6.2, though (sigh). |
---|
1156 | elif (name == "qmaster_response" or name == "element") \ |
---|
1157 | and self.eltq[-1] == "djob_info": # job |
---|
1158 | self.job = {"job_state": "U", "slots": 0, |
---|
1159 | "nodes": [], "queued_timestamp": "", |
---|
1160 | "queued_timestamp": "", "queue": "", |
---|
1161 | "ppn": "0", "RN_max": 0, |
---|
1162 | # fixme in endElement |
---|
1163 | "requested_memory": 0, "requested_time": 0 |
---|
1164 | } |
---|
1165 | self.joblist.append(self.job) |
---|
1166 | elif name == "qstat_l_requests": # resource request |
---|
1167 | self.lrequest = True |
---|
1168 | elif name == "unknown_jobs": |
---|
1169 | raise NoJobs |
---|
1170 | self.eltq.append (name) |
---|
1171 | |
---|
1172 | def characters(self, ch): |
---|
1173 | self.value += ch |
---|
1174 | |
---|
1175 | def endElement(self, name): |
---|
1176 | """Snarf job elements contents into job dictionary. |
---|
1177 | Translate keys if appropriate.""" |
---|
1178 | |
---|
1179 | name_trans = { |
---|
1180 | "JB_job_number": "number", |
---|
1181 | "JB_job_name": "name", "JB_owner": "owner", |
---|
1182 | "queue_name": "queue", "JAT_start_time": "start_timestamp", |
---|
1183 | "JB_submission_time": "queued_timestamp" |
---|
1184 | } |
---|
1185 | value = self.value |
---|
1186 | self.eltq.pop () |
---|
1187 | |
---|
1188 | if name == "djob_info": |
---|
1189 | self.in_joblist = False |
---|
1190 | self.job = {} |
---|
1191 | elif name == "JAT_master_queue": |
---|
1192 | self.job["queue"] = value.split("@")[0] |
---|
1193 | elif name == "JG_qhostname": |
---|
1194 | if not (value in self.job["nodes"]): |
---|
1195 | self.job["nodes"].append(value) |
---|
1196 | elif name == "JG_slots": # slots in use |
---|
1197 | self.job["slots"] += int(value) |
---|
1198 | elif name == "RN_max": # requested slots (tasks or parallel) |
---|
1199 | self.job["RN_max"] = max (self.job["RN_max"], |
---|
1200 | int(value)) |
---|
1201 | elif name == "JAT_state": # job state (bitwise or) |
---|
1202 | value = int (value) |
---|
1203 | # Status values from sge_jobL.h |
---|
1204 | #define JIDLE 0x00000000 |
---|
1205 | #define JHELD 0x00000010 |
---|
1206 | #define JMIGRATING 0x00000020 |
---|
1207 | #define JQUEUED 0x00000040 |
---|
1208 | #define JRUNNING 0x00000080 |
---|
1209 | #define JSUSPENDED 0x00000100 |
---|
1210 | #define JTRANSFERING 0x00000200 |
---|
1211 | #define JDELETED 0x00000400 |
---|
1212 | #define JWAITING 0x00000800 |
---|
1213 | #define JEXITING 0x00001000 |
---|
1214 | #define JWRITTEN 0x00002000 |
---|
1215 | #define JSUSPENDED_ON_THRESHOLD 0x00010000 |
---|
1216 | #define JFINISHED 0x00010000 |
---|
1217 | if value & 0x80: |
---|
1218 | self.job["status"] = "R" |
---|
1219 | elif value & 0x40: |
---|
1220 | self.job["status"] = "Q" |
---|
1221 | else: |
---|
1222 | self.job["status"] = "O" # `other' |
---|
1223 | elif name == "CE_name" and self.lrequest and self.value in \ |
---|
1224 | ("h_cpu", "s_cpu", "cpu", "h_core", "s_core"): |
---|
1225 | # We're in a container for an interesting resource |
---|
1226 | # request; record which type. |
---|
1227 | self.lrequest = self.value |
---|
1228 | elif name == "CE_doubleval" and self.lrequest: |
---|
1229 | # if we're in a container for an interesting |
---|
1230 | # resource request, use the maxmimum of the hard |
---|
1231 | # and soft requests to record the requested CPU |
---|
1232 | # or core. Fixme: I'm not sure if this logic is |
---|
1233 | # right. |
---|
1234 | if self.lrequest in ("h_core", "s_core"): |
---|
1235 | self.job["requested_memory"] = \ |
---|
1236 | max (float (value), |
---|
1237 | self.job["requested_memory"]) |
---|
1238 | # Fixme: Check what cpu means, c.f [hs]_cpu. |
---|
1239 | elif self.lrequest in ("h_cpu", "s_cpu", "cpu"): |
---|
1240 | self.job["requested_time"] = \ |
---|
1241 | max (float (value), |
---|
1242 | self.job["requested_time"]) |
---|
1243 | elif name == "qstat_l_requests": |
---|
1244 | self.lrequest = False |
---|
1245 | elif self.job and self.in_joblist: |
---|
1246 | if name in name_trans: |
---|
1247 | name = name_trans[name] |
---|
1248 | self.job[name] = value |
---|
1249 | |
---|
1250 | # Abstracted from PBS original. |
---|
1251 | # Fixme: Is it worth (or appropriate for PBS) sorting the result? |
---|
1252 | # |
---|
1253 | def do_nodelist( nodes ): |
---|
1254 | |
---|
1255 | """Translate node list as appropriate.""" |
---|
1256 | |
---|
1257 | nodeslist = [ ] |
---|
1258 | my_domain = fqdn_parts( socket.getfqdn() )[1] |
---|
1259 | |
---|
1260 | for node in nodes: |
---|
1261 | |
---|
1262 | host = node.split( '/' )[0] # not relevant for SGE |
---|
1263 | h, host_domain = fqdn_parts(host) |
---|
1264 | |
---|
1265 | if host_domain == my_domain: |
---|
1266 | |
---|
1267 | host = h |
---|
1268 | |
---|
1269 | if nodeslist.count( host ) == 0: |
---|
1270 | |
---|
1271 | for translate_pattern in BATCH_HOST_TRANSLATE: |
---|
1272 | |
---|
1273 | if translate_pattern.find( '/' ) != -1: |
---|
1274 | |
---|
1275 | translate_orig = \ |
---|
1276 | translate_pattern.split( '/' )[1] |
---|
1277 | translate_new = \ |
---|
1278 | translate_pattern.split( '/' )[2] |
---|
1279 | host = re.sub( translate_orig, |
---|
1280 | translate_new, host ) |
---|
1281 | if not host in nodeslist: |
---|
1282 | nodeslist.append( host ) |
---|
1283 | return nodeslist |
---|
1284 | |
---|
1285 | class SLURMDataGatherer( DataGatherer ): |
---|
1286 | |
---|
1287 | global pyslurm |
---|
1288 | |
---|
1289 | """This is the DataGatherer for SLURM""" |
---|
1290 | |
---|
1291 | def __init__( self ): |
---|
1292 | |
---|
1293 | """Setup appropriate variables""" |
---|
1294 | |
---|
1295 | self.jobs = { } |
---|
1296 | self.timeoffset = 0 |
---|
1297 | self.dp = DataProcessor() |
---|
1298 | |
---|
1299 | def getNodeData( self ): |
---|
1300 | |
---|
1301 | slurm_type = pyslurm.node() |
---|
1302 | |
---|
1303 | slurm_nodes = slurm_type.get() |
---|
1304 | |
---|
1305 | nodedict = { } |
---|
1306 | |
---|
1307 | for node, attrs in slurm_nodes.items(): |
---|
1308 | |
---|
1309 | ( num_state, name_state ) = attrs['node_state'] |
---|
1310 | |
---|
1311 | if name_state == 'DOWN': |
---|
1312 | |
---|
1313 | nodedict[ node ] = { 'state' : 'down' } |
---|
1314 | |
---|
1315 | elif name_state == 'DRAIN': |
---|
1316 | |
---|
1317 | nodedict[ node ] = { 'state' : 'offline' } |
---|
1318 | |
---|
1319 | return nodedict |
---|
1320 | |
---|
1321 | def getJobData( self ): |
---|
1322 | |
---|
1323 | """Gather all data on current jobs""" |
---|
1324 | |
---|
1325 | joblist = {} |
---|
1326 | |
---|
1327 | self.cur_time = time.time() |
---|
1328 | |
---|
1329 | slurm_type = pyslurm.job() |
---|
1330 | joblist = slurm_type.get() |
---|
1331 | |
---|
1332 | jobs_processed = [ ] |
---|
1333 | |
---|
1334 | for name, attrs in joblist.items(): |
---|
1335 | display_queue = 1 |
---|
1336 | job_id = name |
---|
1337 | |
---|
1338 | name = self.getAttr( attrs, 'name' ) |
---|
1339 | queue = self.getAttr( attrs, 'partition' ) |
---|
1340 | |
---|
1341 | if QUEUE: |
---|
1342 | for q in QUEUE: |
---|
1343 | if q == queue: |
---|
1344 | display_queue = 1 |
---|
1345 | break |
---|
1346 | else: |
---|
1347 | display_queue = 0 |
---|
1348 | continue |
---|
1349 | if display_queue == 0: |
---|
1350 | continue |
---|
1351 | |
---|
1352 | owner_uid = attrs[ 'user_id' ] |
---|
1353 | ( owner, owner_pw, owner_uid, owner_gid, owner_gecos, owner_dir, owner_shell ) = pwd.getpwuid( owner_uid ) |
---|
1354 | |
---|
1355 | requested_time = self.getAttr( attrs, 'time_limit' ) |
---|
1356 | min_memory = self.getAttr( attrs, 'pn_min_memory' ) |
---|
1357 | |
---|
1358 | if min_memory == 0: |
---|
1359 | |
---|
1360 | requested_memory = '' |
---|
1361 | |
---|
1362 | else: |
---|
1363 | requested_memory = min_memory |
---|
1364 | |
---|
1365 | min_cpus = self.getAttr( attrs, 'pn_min_cpus' ) |
---|
1366 | |
---|
1367 | if min_cpus == 0: |
---|
1368 | |
---|
1369 | ppn = '' |
---|
1370 | |
---|
1371 | else: |
---|
1372 | ppn = min_cpus |
---|
1373 | |
---|
1374 | ( something, status_long ) = self.getAttr( attrs, 'job_state' ) |
---|
1375 | |
---|
1376 | status = 'Q' |
---|
1377 | |
---|
1378 | if status_long == 'RUNNING': |
---|
1379 | |
---|
1380 | status = 'R' |
---|
1381 | |
---|
1382 | elif status_long == 'COMPLETED': |
---|
1383 | |
---|
1384 | continue |
---|
1385 | |
---|
1386 | jobs_processed.append( job_id ) |
---|
1387 | |
---|
1388 | queued_timestamp = self.getAttr( attrs, 'submit_time' ) |
---|
1389 | |
---|
1390 | start_timestamp = '' |
---|
1391 | nodeslist = '' |
---|
1392 | |
---|
1393 | if status == 'R': |
---|
1394 | |
---|
1395 | start_timestamp = self.getAttr( attrs, 'start_time' ) |
---|
1396 | nodes = attrs[ 'nodes' ] |
---|
1397 | |
---|
1398 | if not nodes: |
---|
1399 | |
---|
1400 | # This should not happen |
---|
1401 | |
---|
1402 | # Something wrong: running but 'nodes' returned empty by pyslurm |
---|
1403 | # Possible pyslurm bug: abort/quit/warning |
---|
1404 | |
---|
1405 | err_msg = 'FATAL ERROR: job %s running but nodes returned empty: pyslurm bugged?' %job_id |
---|
1406 | |
---|
1407 | print err_msg |
---|
1408 | debug_msg( 0, err_msg ) |
---|
1409 | sys.exit(1) |
---|
1410 | |
---|
1411 | my_nodelist = [ ] |
---|
1412 | |
---|
1413 | slurm_hostlist = pyslurm.hostlist() |
---|
1414 | slurm_hostlist.create( nodes ) |
---|
1415 | slurm_hostlist.uniq() |
---|
1416 | |
---|
1417 | while slurm_hostlist.count() > 0: |
---|
1418 | |
---|
1419 | my_nodelist.append( slurm_hostlist.pop() ) |
---|
1420 | |
---|
1421 | slurm_hostlist.destroy() |
---|
1422 | |
---|
1423 | del slurm_hostlist |
---|
1424 | |
---|
1425 | nodeslist = do_nodelist( my_nodelist ) |
---|
1426 | |
---|
1427 | if DETECT_TIME_DIFFS: |
---|
1428 | |
---|
1429 | # If a job start if later than our current date, |
---|
1430 | # that must mean the Torque server's time is later |
---|
1431 | # than our local time. |
---|
1432 | |
---|
1433 | if int( start_timestamp ) > int( int( self.cur_time ) + int( self.timeoffset ) ): |
---|
1434 | |
---|
1435 | self.timeoffset = int( int(start_timestamp) - int(self.cur_time) ) |
---|
1436 | |
---|
1437 | elif status == 'Q': |
---|
1438 | |
---|
1439 | nodeslist = str( attrs[ 'num_nodes' ] ) |
---|
1440 | |
---|
1441 | else: |
---|
1442 | start_timestamp = '' |
---|
1443 | nodeslist = '' |
---|
1444 | |
---|
1445 | myAttrs = { } |
---|
1446 | |
---|
1447 | myAttrs[ 'name' ] = str( name ) |
---|
1448 | myAttrs[ 'queue' ] = str( queue ) |
---|
1449 | myAttrs[ 'owner' ] = str( owner ) |
---|
1450 | myAttrs[ 'requested_time' ] = str( requested_time ) |
---|
1451 | myAttrs[ 'requested_memory' ] = str( requested_memory ) |
---|
1452 | myAttrs[ 'ppn' ] = str( ppn ) |
---|
1453 | myAttrs[ 'status' ] = str( status ) |
---|
1454 | myAttrs[ 'start_timestamp' ] = str( start_timestamp ) |
---|
1455 | myAttrs[ 'queued_timestamp' ] = str( queued_timestamp ) |
---|
1456 | myAttrs[ 'reported' ] = str( int( int( self.cur_time ) + int( self.timeoffset ) ) ) |
---|
1457 | myAttrs[ 'nodes' ] = nodeslist |
---|
1458 | myAttrs[ 'domain' ] = fqdn_parts( socket.getfqdn() )[1] |
---|
1459 | myAttrs[ 'poll_interval' ] = str( BATCH_POLL_INTERVAL ) |
---|
1460 | |
---|
1461 | if self.jobDataChanged( self.jobs, job_id, myAttrs ) and myAttrs['status'] in [ 'R', 'Q' ]: |
---|
1462 | |
---|
1463 | self.jobs[ job_id ] = myAttrs |
---|
1464 | |
---|
1465 | for id, attrs in self.jobs.items(): |
---|
1466 | |
---|
1467 | if id not in jobs_processed: |
---|
1468 | |
---|
1469 | # This one isn't there anymore; toedeledoki! |
---|
1470 | # |
---|
1471 | del self.jobs[ id ] |
---|
1472 | |
---|
1473 | class SgeDataGatherer(DataGatherer): |
---|
1474 | |
---|
1475 | jobs = {} |
---|
1476 | |
---|
1477 | def __init__( self ): |
---|
1478 | self.jobs = {} |
---|
1479 | self.timeoffset = 0 |
---|
1480 | self.dp = DataProcessor() |
---|
1481 | |
---|
1482 | def getJobData( self ): |
---|
1483 | """Gather all data on current jobs in SGE""" |
---|
1484 | |
---|
1485 | import popen2 |
---|
1486 | |
---|
1487 | self.cur_time = 0 |
---|
1488 | queues = "" |
---|
1489 | if QUEUE: # only for specific queues |
---|
1490 | # Fixme: assumes queue names don't contain single |
---|
1491 | # quote or comma. Don't know what the SGE rules are. |
---|
1492 | queues = " -q '" + string.join (QUEUE, ",") + "'" |
---|
1493 | # Note the comment in SgeQstatXMLParser about scaling with |
---|
1494 | # this method of getting data. I haven't found better one. |
---|
1495 | # Output with args `-xml -ext -f -r' is easier to parse |
---|
1496 | # in some ways, harder in others, but it doesn't provide |
---|
1497 | # the submission time (at least SGE 6.0). The pipeline |
---|
1498 | # into sed corrects bogus XML observed with a configuration |
---|
1499 | # of SGE 6.0u8, which otherwise causes the parsing to hang. |
---|
1500 | piping = popen2.Popen3("qstat -u '*' -j '*' -xml | \ |
---|
1501 | sed -e 's/reported usage>/reported_usage>/g' -e 's;<\/*JATASK:.*>;;'" \ |
---|
1502 | + queues, True) |
---|
1503 | qstatparser = SgeQstatXMLParser() |
---|
1504 | parse_err = 0 |
---|
1505 | try: |
---|
1506 | xml.sax.parse(piping.fromchild, qstatparser) |
---|
1507 | except NoJobs: |
---|
1508 | pass |
---|
1509 | except: |
---|
1510 | parse_err = 1 |
---|
1511 | if piping.wait(): |
---|
1512 | debug_msg(10, "qstat error, skipping until next polling interval: " + piping.childerr.readline()) |
---|
1513 | return None |
---|
1514 | elif parse_err: |
---|
1515 | debug_msg(10, "Bad XML output from qstat"()) |
---|
1516 | exit (1) |
---|
1517 | for f in piping.fromchild, piping.tochild, piping.childerr: |
---|
1518 | f.close() |
---|
1519 | self.cur_time = time.time() |
---|
1520 | jobs_processed = [] |
---|
1521 | for job in qstatparser.joblist: |
---|
1522 | job_id = job["number"] |
---|
1523 | if job["status"] in [ 'Q', 'R' ]: |
---|
1524 | jobs_processed.append(job_id) |
---|
1525 | if job["status"] == "R": |
---|
1526 | job["nodes"] = do_nodelist (job["nodes"]) |
---|
1527 | # Fixme: why is job["nodes"] sometimes null? |
---|
1528 | try: |
---|
1529 | # Fixme: Is this sensible? The |
---|
1530 | # PBS-type PPN isn't something you use |
---|
1531 | # with SGE. |
---|
1532 | job["ppn"] = float(job["slots"]) / len(job["nodes"]) |
---|
1533 | except: |
---|
1534 | job["ppn"] = 0 |
---|
1535 | if DETECT_TIME_DIFFS: |
---|
1536 | # If a job start is later than our |
---|
1537 | # current date, that must mean |
---|
1538 | # the SGE server's time is later |
---|
1539 | # than our local time. |
---|
1540 | start_timestamp = int (job["start_timestamp"]) |
---|
1541 | if start_timestamp > int(self.cur_time) + int(self.timeoffset): |
---|
1542 | |
---|
1543 | self.timeoffset = start_timestamp - int(self.cur_time) |
---|
1544 | else: |
---|
1545 | # fixme: Note sure what this should be: |
---|
1546 | job["ppn"] = job["RN_max"] |
---|
1547 | job["nodes"] = "1" |
---|
1548 | |
---|
1549 | myAttrs = {} |
---|
1550 | for attr in ["name", "queue", "owner", |
---|
1551 | "requested_time", "status", |
---|
1552 | "requested_memory", "ppn", |
---|
1553 | "start_timestamp", "queued_timestamp"]: |
---|
1554 | myAttrs[attr] = str(job[attr]) |
---|
1555 | myAttrs["nodes"] = job["nodes"] |
---|
1556 | myAttrs["reported"] = str(int(self.cur_time) + int(self.timeoffset)) |
---|
1557 | myAttrs["domain"] = fqdn_parts(socket.getfqdn())[1] |
---|
1558 | myAttrs["poll_interval"] = str(BATCH_POLL_INTERVAL) |
---|
1559 | |
---|
1560 | if self.jobDataChanged(self.jobs, job_id, myAttrs) and myAttrs["status"] in ["R", "Q"]: |
---|
1561 | self.jobs[job_id] = myAttrs |
---|
1562 | for id, attrs in self.jobs.items(): |
---|
1563 | if id not in jobs_processed: |
---|
1564 | del self.jobs[id] |
---|
1565 | |
---|
1566 | # LSF code by Mahmoud Hanafi <hanafim@users.sourceforge.nt> |
---|
1567 | # Requres LSFObject http://sourceforge.net/projects/lsfobject |
---|
1568 | # |
---|
1569 | class LsfDataGatherer(DataGatherer): |
---|
1570 | |
---|
1571 | """This is the DataGatherer for LSf""" |
---|
1572 | |
---|
1573 | global lsfObject |
---|
1574 | |
---|
1575 | def __init__( self ): |
---|
1576 | |
---|
1577 | self.jobs = { } |
---|
1578 | self.timeoffset = 0 |
---|
1579 | self.dp = DataProcessor() |
---|
1580 | self.initLsfQuery() |
---|
1581 | |
---|
1582 | def _countDuplicatesInList( self, dupedList ): |
---|
1583 | |
---|
1584 | countDupes = { } |
---|
1585 | |
---|
1586 | for item in dupedList: |
---|
1587 | |
---|
1588 | if not countDupes.has_key( item ): |
---|
1589 | |
---|
1590 | countDupes[ item ] = 1 |
---|
1591 | else: |
---|
1592 | countDupes[ item ] = countDupes[ item ] + 1 |
---|
1593 | |
---|
1594 | dupeCountList = [ ] |
---|
1595 | |
---|
1596 | for item, count in countDupes.items(): |
---|
1597 | |
---|
1598 | dupeCountList.append( ( item, count ) ) |
---|
1599 | |
---|
1600 | return dupeCountList |
---|
1601 | # |
---|
1602 | #lst = ['I1','I2','I1','I3','I4','I4','I7','I7','I7','I7','I7'] |
---|
1603 | #print _countDuplicatesInList(lst) |
---|
1604 | #[('I1', 2), ('I3', 1), ('I2', 1), ('I4', 2), ('I7', 5)] |
---|
1605 | ######################## |
---|
1606 | |
---|
1607 | def initLsfQuery( self ): |
---|
1608 | self.pq = None |
---|
1609 | self.pq = lsfObject.jobInfoEntObject() |
---|
1610 | |
---|
1611 | def getJobData( self, known_jobs="" ): |
---|
1612 | """Gather all data on current jobs in LSF""" |
---|
1613 | if len( known_jobs ) > 0: |
---|
1614 | jobs = known_jobs |
---|
1615 | else: |
---|
1616 | jobs = { } |
---|
1617 | joblist = {} |
---|
1618 | joblist = self.pq.getJobInfo() |
---|
1619 | nodelist = '' |
---|
1620 | |
---|
1621 | self.cur_time = time.time() |
---|
1622 | |
---|
1623 | jobs_processed = [ ] |
---|
1624 | |
---|
1625 | for name, attrs in joblist.items(): |
---|
1626 | job_id = str(name) |
---|
1627 | jobs_processed.append( job_id ) |
---|
1628 | name = self.getAttr( attrs, 'jobName' ) |
---|
1629 | queue = self.getAttr( self.getAttr( attrs, 'submit') , 'queue' ) |
---|
1630 | owner = self.getAttr( attrs, 'user' ) |
---|
1631 | |
---|
1632 | ### THIS IS THE rLimit List index values |
---|
1633 | #define LSF_RLIMIT_CPU 0 /* cpu time in milliseconds */ |
---|
1634 | #define LSF_RLIMIT_FSIZE 1 /* maximum file size */ |
---|
1635 | #define LSF_RLIMIT_DATA 2 /* data size */ |
---|
1636 | #define LSF_RLIMIT_STACK 3 /* stack size */ |
---|
1637 | #define LSF_RLIMIT_CORE 4 /* core file size */ |
---|
1638 | #define LSF_RLIMIT_RSS 5 /* resident set size */ |
---|
1639 | #define LSF_RLIMIT_NOFILE 6 /* open files */ |
---|
1640 | #define LSF_RLIMIT_OPEN_MAX 7 /* (from HP-UX) */ |
---|
1641 | #define LSF_RLIMIT_VMEM 8 /* maximum swap mem */ |
---|
1642 | #define LSF_RLIMIT_SWAP 8 |
---|
1643 | #define LSF_RLIMIT_RUN 9 /* max wall-clock time limit */ |
---|
1644 | #define LSF_RLIMIT_PROCESS 10 /* process number limit */ |
---|
1645 | #define LSF_RLIMIT_THREAD 11 /* thread number limit (introduced in LSF6.0) */ |
---|
1646 | #define LSF_RLIM_NLIMITS 12 /* number of resource limits */ |
---|
1647 | |
---|
1648 | requested_time = self.getAttr( self.getAttr( attrs, 'submit') , 'rLimits' )[9] |
---|
1649 | if requested_time == -1: |
---|
1650 | requested_time = "" |
---|
1651 | requested_memory = self.getAttr( self.getAttr( attrs, 'submit') , 'rLimits' )[8] |
---|
1652 | if requested_memory == -1: |
---|
1653 | requested_memory = "" |
---|
1654 | # This tries to get proc per node. We don't support this right now |
---|
1655 | ppn = 0 #self.getAttr( self.getAttr( attrs, 'SubmitList') , 'numProessors' ) |
---|
1656 | requested_cpus = self.getAttr( self.getAttr( attrs, 'submit') , 'numProcessors' ) |
---|
1657 | if requested_cpus == None or requested_cpus == "": |
---|
1658 | requested_cpus = 1 |
---|
1659 | |
---|
1660 | if QUEUE: |
---|
1661 | for q in QUEUE: |
---|
1662 | if q == queue: |
---|
1663 | display_queue = 1 |
---|
1664 | break |
---|
1665 | else: |
---|
1666 | display_queue = 0 |
---|
1667 | continue |
---|
1668 | if display_queue == 0: |
---|
1669 | continue |
---|
1670 | |
---|
1671 | runState = self.getAttr( attrs, 'status' ) |
---|
1672 | if runState == 4: |
---|
1673 | status = 'R' |
---|
1674 | else: |
---|
1675 | status = 'Q' |
---|
1676 | queued_timestamp = self.getAttr( attrs, 'submitTime' ) |
---|
1677 | |
---|
1678 | if status == 'R': |
---|
1679 | start_timestamp = self.getAttr( attrs, 'startTime' ) |
---|
1680 | nodesCpu = dict(self._countDuplicatesInList(self.getAttr( attrs, 'exHosts' ))) |
---|
1681 | nodelist = nodesCpu.keys() |
---|
1682 | |
---|
1683 | if DETECT_TIME_DIFFS: |
---|
1684 | |
---|
1685 | # If a job start if later than our current date, |
---|
1686 | # that must mean the Torque server's time is later |
---|
1687 | # than our local time. |
---|
1688 | |
---|
1689 | if int(start_timestamp) > int( int(self.cur_time) + int(self.timeoffset) ): |
---|
1690 | |
---|
1691 | self.timeoffset = int( int(start_timestamp) - int(self.cur_time) ) |
---|
1692 | |
---|
1693 | elif status == 'Q': |
---|
1694 | start_timestamp = '' |
---|
1695 | count_mynodes = 0 |
---|
1696 | numeric_node = 1 |
---|
1697 | nodelist = '' |
---|
1698 | |
---|
1699 | myAttrs = { } |
---|
1700 | if name == "": |
---|
1701 | myAttrs['name'] = "none" |
---|
1702 | else: |
---|
1703 | myAttrs['name'] = name |
---|
1704 | |
---|
1705 | myAttrs[ 'owner' ] = owner |
---|
1706 | myAttrs[ 'requested_time' ] = str(requested_time) |
---|
1707 | myAttrs[ 'requested_memory' ] = str(requested_memory) |
---|
1708 | myAttrs[ 'requested_cpus' ] = str(requested_cpus) |
---|
1709 | myAttrs[ 'ppn' ] = str( ppn ) |
---|
1710 | myAttrs[ 'status' ] = status |
---|
1711 | myAttrs[ 'start_timestamp' ] = str(start_timestamp) |
---|
1712 | myAttrs[ 'queue' ] = str(queue) |
---|
1713 | myAttrs[ 'queued_timestamp' ] = str(queued_timestamp) |
---|
1714 | myAttrs[ 'reported' ] = str( int( int( self.cur_time ) + int( self.timeoffset ) ) ) |
---|
1715 | myAttrs[ 'nodes' ] = do_nodelist( nodelist ) |
---|
1716 | myAttrs[ 'domain' ] = fqdn_parts( socket.getfqdn() )[1] |
---|
1717 | myAttrs[ 'poll_interval' ] = str(BATCH_POLL_INTERVAL) |
---|
1718 | |
---|
1719 | if self.jobDataChanged( jobs, job_id, myAttrs ) and myAttrs['status'] in [ 'R', 'Q' ]: |
---|
1720 | jobs[ job_id ] = myAttrs |
---|
1721 | |
---|
1722 | debug_msg( 10, printTime() + ' job %s state changed' %(job_id) ) |
---|
1723 | |
---|
1724 | for id, attrs in jobs.items(): |
---|
1725 | if id not in jobs_processed: |
---|
1726 | # This one isn't there anymore |
---|
1727 | # |
---|
1728 | del jobs[ id ] |
---|
1729 | self.jobs=jobs |
---|
1730 | |
---|
1731 | |
---|
1732 | class PbsDataGatherer( DataGatherer ): |
---|
1733 | |
---|
1734 | """This is the DataGatherer for PBS and Torque""" |
---|
1735 | |
---|
1736 | global PBSQuery, PBSError |
---|
1737 | |
---|
1738 | def __init__( self ): |
---|
1739 | |
---|
1740 | """Setup appropriate variables""" |
---|
1741 | |
---|
1742 | self.jobs = { } |
---|
1743 | self.timeoffset = 0 |
---|
1744 | self.dp = DataProcessor() |
---|
1745 | |
---|
1746 | self.initPbsQuery() |
---|
1747 | |
---|
1748 | def initPbsQuery( self ): |
---|
1749 | |
---|
1750 | self.pq = None |
---|
1751 | |
---|
1752 | try: |
---|
1753 | |
---|
1754 | if( BATCH_SERVER ): |
---|
1755 | |
---|
1756 | self.pq = PBSQuery( BATCH_SERVER ) |
---|
1757 | else: |
---|
1758 | self.pq = PBSQuery() |
---|
1759 | |
---|
1760 | except PBSError, details: |
---|
1761 | print 'Cannot connect to pbs server' |
---|
1762 | print details |
---|
1763 | sys.exit( 1 ) |
---|
1764 | |
---|
1765 | try: |
---|
1766 | # TODO: actually use new data structure |
---|
1767 | self.pq.old_data_structure() |
---|
1768 | |
---|
1769 | except AttributeError: |
---|
1770 | |
---|
1771 | # pbs_query is older |
---|
1772 | # |
---|
1773 | pass |
---|
1774 | |
---|
1775 | def getNodeData( self ): |
---|
1776 | |
---|
1777 | nodedict = { } |
---|
1778 | |
---|
1779 | try: |
---|
1780 | nodedict = self.pq.getnodes() |
---|
1781 | |
---|
1782 | except PBSError, detail: |
---|
1783 | |
---|
1784 | debug_msg( 10, "PBS server unavailable, skipping until next polling interval: " + str( detail ) ) |
---|
1785 | |
---|
1786 | return nodedict |
---|
1787 | |
---|
1788 | def getJobData( self ): |
---|
1789 | |
---|
1790 | """Gather all data on current jobs in Torque""" |
---|
1791 | |
---|
1792 | joblist = {} |
---|
1793 | self.cur_time = 0 |
---|
1794 | |
---|
1795 | try: |
---|
1796 | joblist = self.pq.getjobs() |
---|
1797 | self.cur_time = time.time() |
---|
1798 | |
---|
1799 | except (PBSError, TypeError), detail: |
---|
1800 | |
---|
1801 | debug_msg( 10, "PBS server unavailable, skipping until next polling interval: " + str( detail ) ) |
---|
1802 | return None |
---|
1803 | |
---|
1804 | jobs_processed = [ ] |
---|
1805 | |
---|
1806 | for name, attrs in joblist.items(): |
---|
1807 | display_queue = 1 |
---|
1808 | job_id = name.split( '.' )[0] |
---|
1809 | |
---|
1810 | name = self.getAttr( attrs, 'Job_Name' ) |
---|
1811 | queue = self.getAttr( attrs, 'queue' ) |
---|
1812 | |
---|
1813 | if QUEUE: |
---|
1814 | for q in QUEUE: |
---|
1815 | if q == queue: |
---|
1816 | display_queue = 1 |
---|
1817 | break |
---|
1818 | else: |
---|
1819 | display_queue = 0 |
---|
1820 | continue |
---|
1821 | if display_queue == 0: |
---|
1822 | continue |
---|
1823 | |
---|
1824 | |
---|
1825 | owner = self.getAttr( attrs, 'Job_Owner' ).split( '@' )[0] |
---|
1826 | requested_time = self.getAttr( attrs, 'Resource_List.walltime' ) |
---|
1827 | requested_memory = self.getAttr( attrs, 'Resource_List.mem' ) |
---|
1828 | |
---|
1829 | mynoderequest = self.getAttr( attrs, 'Resource_List.nodes' ) |
---|
1830 | |
---|
1831 | ppn = '' |
---|
1832 | |
---|
1833 | if mynoderequest.find( ':' ) != -1 and mynoderequest.find( 'ppn' ) != -1: |
---|
1834 | |
---|
1835 | mynoderequest_fields = mynoderequest.split( ':' ) |
---|
1836 | |
---|
1837 | for mynoderequest_field in mynoderequest_fields: |
---|
1838 | |
---|
1839 | if mynoderequest_field.find( 'ppn' ) != -1: |
---|
1840 | |
---|
1841 | ppn = mynoderequest_field.split( 'ppn=' )[1] |
---|
1842 | |
---|
1843 | status = self.getAttr( attrs, 'job_state' ) |
---|
1844 | |
---|
1845 | if status in [ 'Q', 'R' ]: |
---|
1846 | |
---|
1847 | jobs_processed.append( job_id ) |
---|
1848 | |
---|
1849 | queued_timestamp = self.getAttr( attrs, 'ctime' ) |
---|
1850 | |
---|
1851 | if status == 'R': |
---|
1852 | |
---|
1853 | start_timestamp = self.getAttr( attrs, 'mtime' ) |
---|
1854 | nodes = self.getAttr( attrs, 'exec_host' ).split( '+' ) |
---|
1855 | |
---|
1856 | nodeslist = do_nodelist( nodes ) |
---|
1857 | |
---|
1858 | if DETECT_TIME_DIFFS: |
---|
1859 | |
---|
1860 | # If a job start if later than our current date, |
---|
1861 | # that must mean the Torque server's time is later |
---|
1862 | # than our local time. |
---|
1863 | |
---|
1864 | if int( start_timestamp ) > int( int( self.cur_time ) + int( self.timeoffset ) ): |
---|
1865 | |
---|
1866 | self.timeoffset = int( int(start_timestamp) - int(self.cur_time) ) |
---|
1867 | |
---|
1868 | elif status == 'Q': |
---|
1869 | |
---|
1870 | # 'mynodequest' can be a string in the following syntax according to the |
---|
1871 | # Torque Administator's manual: |
---|
1872 | # |
---|
1873 | # {<node_count> | <hostname>}[:ppn=<ppn>][:<property>[:<property>]...][+ ...] |
---|
1874 | # {<node_count> | <hostname>}[:ppn=<ppn>][:<property>[:<property>]...][+ ...] |
---|
1875 | # etc |
---|
1876 | # |
---|
1877 | |
---|
1878 | # |
---|
1879 | # For now we only count the amount of nodes request and ignore properties |
---|
1880 | # |
---|
1881 | |
---|
1882 | start_timestamp = '' |
---|
1883 | count_mynodes = 0 |
---|
1884 | |
---|
1885 | for node in mynoderequest.split( '+' ): |
---|
1886 | |
---|
1887 | # Just grab the {node_count|hostname} part and ignore properties |
---|
1888 | # |
---|
1889 | nodepart = node.split( ':' )[0] |
---|
1890 | |
---|
1891 | # Let's assume a node_count value |
---|
1892 | # |
---|
1893 | numeric_node = 1 |
---|
1894 | |
---|
1895 | # Chop the value up into characters |
---|
1896 | # |
---|
1897 | for letter in nodepart: |
---|
1898 | |
---|
1899 | # If this char is not a digit (0-9), this must be a hostname |
---|
1900 | # |
---|
1901 | if letter not in string.digits: |
---|
1902 | |
---|
1903 | numeric_node = 0 |
---|
1904 | |
---|
1905 | # If this is a hostname, just count this as one (1) node |
---|
1906 | # |
---|
1907 | if not numeric_node: |
---|
1908 | |
---|
1909 | count_mynodes = count_mynodes + 1 |
---|
1910 | else: |
---|
1911 | |
---|
1912 | # If this a number, it must be the node_count |
---|
1913 | # and increase our count with it's value |
---|
1914 | # |
---|
1915 | try: |
---|
1916 | count_mynodes = count_mynodes + int( nodepart ) |
---|
1917 | |
---|
1918 | except ValueError, detail: |
---|
1919 | |
---|
1920 | # When we arrive here I must be bugged or very confused |
---|
1921 | # THIS SHOULD NOT HAPPEN! |
---|
1922 | # |
---|
1923 | debug_msg( 10, str( detail ) ) |
---|
1924 | debug_msg( 10, "Encountered weird node in Resources_List?!" ) |
---|
1925 | debug_msg( 10, 'nodepart = ' + str( nodepart ) ) |
---|
1926 | debug_msg( 10, 'job = ' + str( name ) ) |
---|
1927 | debug_msg( 10, 'attrs = ' + str( attrs ) ) |
---|
1928 | |
---|
1929 | nodeslist = str( count_mynodes ) |
---|
1930 | else: |
---|
1931 | start_timestamp = '' |
---|
1932 | nodeslist = '' |
---|
1933 | |
---|
1934 | myAttrs = { } |
---|
1935 | |
---|
1936 | myAttrs[ 'name' ] = str( name ) |
---|
1937 | myAttrs[ 'queue' ] = str( queue ) |
---|
1938 | myAttrs[ 'owner' ] = str( owner ) |
---|
1939 | myAttrs[ 'requested_time' ] = str( requested_time ) |
---|
1940 | myAttrs[ 'requested_memory' ] = str( requested_memory ) |
---|
1941 | myAttrs[ 'ppn' ] = str( ppn ) |
---|
1942 | myAttrs[ 'status' ] = str( status ) |
---|
1943 | myAttrs[ 'start_timestamp' ] = str( start_timestamp ) |
---|
1944 | myAttrs[ 'queued_timestamp' ] = str( queued_timestamp ) |
---|
1945 | myAttrs[ 'reported' ] = str( int( int( self.cur_time ) + int( self.timeoffset ) ) ) |
---|
1946 | myAttrs[ 'nodes' ] = nodeslist |
---|
1947 | myAttrs[ 'domain' ] = fqdn_parts( socket.getfqdn() )[1] |
---|
1948 | myAttrs[ 'poll_interval' ] = str( BATCH_POLL_INTERVAL ) |
---|
1949 | |
---|
1950 | if self.jobDataChanged( self.jobs, job_id, myAttrs ) and myAttrs['status'] in [ 'R', 'Q' ]: |
---|
1951 | |
---|
1952 | self.jobs[ job_id ] = myAttrs |
---|
1953 | |
---|
1954 | for id, attrs in self.jobs.items(): |
---|
1955 | |
---|
1956 | if id not in jobs_processed: |
---|
1957 | |
---|
1958 | # This one isn't there anymore; toedeledoki! |
---|
1959 | # |
---|
1960 | del self.jobs[ id ] |
---|
1961 | |
---|
1962 | GMETRIC_DEFAULT_TYPE = 'string' |
---|
1963 | GMETRIC_DEFAULT_HOST = '127.0.0.1' |
---|
1964 | GMETRIC_DEFAULT_PORT = '8649' |
---|
1965 | GMETRIC_DEFAULT_UNITS = '' |
---|
1966 | |
---|
1967 | class Gmetric: |
---|
1968 | |
---|
1969 | global GMETRIC_DEFAULT_HOST, GMETRIC_DEFAULT_PORT |
---|
1970 | |
---|
1971 | slope = { 'zero' : 0, 'positive' : 1, 'negative' : 2, 'both' : 3, 'unspecified' : 4 } |
---|
1972 | type = ( '', 'string', 'uint16', 'int16', 'uint32', 'int32', 'float', 'double', 'timestamp' ) |
---|
1973 | protocol = ( 'udp', 'multicast' ) |
---|
1974 | |
---|
1975 | def __init__( self, host=GMETRIC_DEFAULT_HOST, port=GMETRIC_DEFAULT_PORT ): |
---|
1976 | |
---|
1977 | global GMETRIC_DEFAULT_TYPE |
---|
1978 | |
---|
1979 | self.prot = self.checkHostProtocol( host ) |
---|
1980 | self.data_msg = xdrlib.Packer() |
---|
1981 | self.meta_msg = xdrlib.Packer() |
---|
1982 | self.socket = socket.socket( socket.AF_INET, socket.SOCK_DGRAM ) |
---|
1983 | |
---|
1984 | if self.prot not in self.protocol: |
---|
1985 | |
---|
1986 | raise ValueError( "Protocol must be one of: " + str( self.protocol ) ) |
---|
1987 | |
---|
1988 | if self.prot == 'multicast': |
---|
1989 | |
---|
1990 | # Set multicast options |
---|
1991 | # |
---|
1992 | self.socket.setsockopt( socket.IPPROTO_IP, socket.IP_MULTICAST_TTL, 20 ) |
---|
1993 | |
---|
1994 | self.hostport = ( host, int( port ) ) |
---|
1995 | self.slopestr = 'both' |
---|
1996 | self.tmax = 60 |
---|
1997 | |
---|
1998 | def __del__( self ): |
---|
1999 | |
---|
2000 | self.socket.close() |
---|
2001 | |
---|
2002 | def checkHostProtocol( self, ip ): |
---|
2003 | |
---|
2004 | """Detect if a ip adress is a multicast address""" |
---|
2005 | |
---|
2006 | MULTICAST_ADDRESS_MIN = ( "224", "0", "0", "0" ) |
---|
2007 | MULTICAST_ADDRESS_MAX = ( "239", "255", "255", "255" ) |
---|
2008 | |
---|
2009 | ip_fields = ip.split( '.' ) |
---|
2010 | |
---|
2011 | if ip_fields >= MULTICAST_ADDRESS_MIN and ip_fields <= MULTICAST_ADDRESS_MAX: |
---|
2012 | |
---|
2013 | return 'multicast' |
---|
2014 | else: |
---|
2015 | return 'udp' |
---|
2016 | |
---|
2017 | def send( self, name, value, dmax, typestr = '', units = '' ): |
---|
2018 | |
---|
2019 | if len( units ) == 0: |
---|
2020 | units = GMETRIC_DEFAULT_UNITS |
---|
2021 | |
---|
2022 | if len( typestr ) == 0: |
---|
2023 | typestr = GMETRIC_DEFAULT_TYPE |
---|
2024 | |
---|
2025 | (meta_msg, data_msg) = self.makexdr( name, value, typestr, units, self.slopestr, self.tmax, dmax ) |
---|
2026 | |
---|
2027 | meta_rt = self.socket.sendto( meta_msg, self.hostport ) |
---|
2028 | data_rt = self.socket.sendto( data_msg, self.hostport ) |
---|
2029 | |
---|
2030 | return ( meta_rt, data_rt ) |
---|
2031 | |
---|
2032 | def makexdr( self, name, value, typestr, unitstr, slopestr, tmax, dmax, group=None, spoof=None ): |
---|
2033 | |
---|
2034 | hostname = "unset" |
---|
2035 | |
---|
2036 | if slopestr not in self.slope: |
---|
2037 | |
---|
2038 | raise ValueError( "Slope must be one of: " + str( self.slope.keys() ) ) |
---|
2039 | |
---|
2040 | if typestr not in self.type: |
---|
2041 | |
---|
2042 | raise ValueError( "Type must be one of: " + str( self.type ) ) |
---|
2043 | |
---|
2044 | if len( name ) == 0: |
---|
2045 | |
---|
2046 | raise ValueError( "Name must be non-empty" ) |
---|
2047 | |
---|
2048 | self.meta_msg.reset() |
---|
2049 | self.meta_msg.pack_int( 128 ) |
---|
2050 | |
---|
2051 | if not spoof: |
---|
2052 | self.meta_msg.pack_string( hostname ) |
---|
2053 | else: |
---|
2054 | self.meta_msg.pack_string( spoof ) |
---|
2055 | |
---|
2056 | self.meta_msg.pack_string( name ) |
---|
2057 | |
---|
2058 | if not spoof: |
---|
2059 | self.meta_msg.pack_int( 0 ) |
---|
2060 | else: |
---|
2061 | self.meta_msg.pack_int( 1 ) |
---|
2062 | |
---|
2063 | self.meta_msg.pack_string( typestr ) |
---|
2064 | self.meta_msg.pack_string( name ) |
---|
2065 | self.meta_msg.pack_string( unitstr ) |
---|
2066 | self.meta_msg.pack_int( self.slope[ slopestr ] ) |
---|
2067 | self.meta_msg.pack_uint( int( tmax ) ) |
---|
2068 | self.meta_msg.pack_uint( int( dmax ) ) |
---|
2069 | |
---|
2070 | if not group: |
---|
2071 | self.meta_msg.pack_int( 0 ) |
---|
2072 | else: |
---|
2073 | self.meta_msg.pack_int( 1 ) |
---|
2074 | self.meta_msg.pack_string( "GROUP" ) |
---|
2075 | self.meta_msg.pack_string( group ) |
---|
2076 | |
---|
2077 | self.data_msg.reset() |
---|
2078 | self.data_msg.pack_int( 128+5 ) |
---|
2079 | |
---|
2080 | if not spoof: |
---|
2081 | self.data_msg.pack_string( hostname ) |
---|
2082 | else: |
---|
2083 | self.data_msg.pack_string( spoof ) |
---|
2084 | |
---|
2085 | self.data_msg.pack_string( name ) |
---|
2086 | |
---|
2087 | if not spoof: |
---|
2088 | self.data_msg.pack_int( 0 ) |
---|
2089 | else: |
---|
2090 | self.data_msg.pack_int( 1 ) |
---|
2091 | |
---|
2092 | self.data_msg.pack_string( "%s" ) |
---|
2093 | self.data_msg.pack_string( str( value ) ) |
---|
2094 | |
---|
2095 | return ( self.meta_msg.get_buffer(), self.data_msg.get_buffer() ) |
---|
2096 | |
---|
2097 | def printTime( ): |
---|
2098 | |
---|
2099 | """Print current time/date in human readable format for log/debug""" |
---|
2100 | |
---|
2101 | return time.strftime("%a, %d %b %Y %H:%M:%S") |
---|
2102 | |
---|
2103 | def debug_msg( level, msg ): |
---|
2104 | |
---|
2105 | """Print msg if at or above current debug level""" |
---|
2106 | |
---|
2107 | global DAEMONIZE, DEBUG_LEVEL, SYSLOG_LEVEL |
---|
2108 | |
---|
2109 | if (not DAEMONIZE and DEBUG_LEVEL >= level): |
---|
2110 | sys.stderr.write( msg + '\n' ) |
---|
2111 | |
---|
2112 | if (DAEMONIZE and USE_SYSLOG and SYSLOG_LEVEL >= level): |
---|
2113 | syslog.syslog( msg ) |
---|
2114 | |
---|
2115 | def write_pidfile(): |
---|
2116 | |
---|
2117 | # Write pidfile if PIDFILE is set |
---|
2118 | # |
---|
2119 | if PIDFILE: |
---|
2120 | |
---|
2121 | pid = os.getpid() |
---|
2122 | |
---|
2123 | pidfile = open( PIDFILE, 'w' ) |
---|
2124 | |
---|
2125 | pidfile.write( str( pid ) ) |
---|
2126 | pidfile.close() |
---|
2127 | |
---|
2128 | def main(): |
---|
2129 | |
---|
2130 | """Application start""" |
---|
2131 | |
---|
2132 | global PBSQuery, PBSError, lsfObject, pyslurm |
---|
2133 | global SYSLOG_FACILITY, USE_SYSLOG, BATCH_API, DAEMONIZE, BATCH_SERVER |
---|
2134 | |
---|
2135 | if not processArgs( sys.argv[1:] ): |
---|
2136 | |
---|
2137 | sys.exit( 1 ) |
---|
2138 | |
---|
2139 | # Load appropriate DataGatherer depending on which BATCH_API is set |
---|
2140 | # and any required modules for the Gatherer |
---|
2141 | # |
---|
2142 | if BATCH_API == 'pbs': |
---|
2143 | |
---|
2144 | try: |
---|
2145 | from PBSQuery import PBSQuery, PBSError |
---|
2146 | |
---|
2147 | except ImportError, details: |
---|
2148 | |
---|
2149 | print "FATAL ERROR: BATCH_API set to 'pbs' but python module 'pbs_python' is not found or installed" |
---|
2150 | print details |
---|
2151 | sys.exit( 1 ) |
---|
2152 | |
---|
2153 | gather = PbsDataGatherer() |
---|
2154 | |
---|
2155 | elif BATCH_API == 'sge': |
---|
2156 | |
---|
2157 | if BATCH_SERVER != 'localhost': |
---|
2158 | |
---|
2159 | # Print and log, but continue execution |
---|
2160 | err_msg = "WARNING: BATCH_API 'sge' ignores BATCH_SERVER (can only be 'localhost')" |
---|
2161 | print err_msg |
---|
2162 | debug_msg( 0, err_msg ) |
---|
2163 | |
---|
2164 | # Tested with SGE 6.0u11. |
---|
2165 | # |
---|
2166 | gather = SgeDataGatherer() |
---|
2167 | |
---|
2168 | elif BATCH_API == 'lsf': |
---|
2169 | |
---|
2170 | if BATCH_SERVER != 'localhost': |
---|
2171 | |
---|
2172 | # Print and log, but continue execution |
---|
2173 | err_msg = "WARNING: BATCH_API 'lsf' ignores BATCH_SERVER (can only be 'localhost')" |
---|
2174 | print err_msg |
---|
2175 | debug_msg( 0, err_msg ) |
---|
2176 | |
---|
2177 | try: |
---|
2178 | from lsfObject import lsfObject |
---|
2179 | except: |
---|
2180 | print "FATAL ERROR: BATCH_API set to 'lsf' but python module 'lsfObject' is not found or installed" |
---|
2181 | sys.exit( 1 ) |
---|
2182 | |
---|
2183 | gather = LsfDataGatherer() |
---|
2184 | |
---|
2185 | elif BATCH_API == 'slurm': |
---|
2186 | |
---|
2187 | if BATCH_SERVER != 'localhost': |
---|
2188 | |
---|
2189 | # Print and log, but continue execution |
---|
2190 | err_msg = "WARNING: BATCH_API 'slurm' ignores BATCH_SERVER (can only be 'localhost')" |
---|
2191 | print err_msg |
---|
2192 | debug_msg( 0, err_msg ) |
---|
2193 | |
---|
2194 | try: |
---|
2195 | import pyslurm |
---|
2196 | except: |
---|
2197 | print "FATAL ERROR: BATCH_API set to 'slurm' but python module is not found or installed" |
---|
2198 | sys.exit( 1 ) |
---|
2199 | |
---|
2200 | gather = SLURMDataGatherer() |
---|
2201 | |
---|
2202 | else: |
---|
2203 | print "FATAL ERROR: unknown BATCH_API '" + BATCH_API + "' is not supported" |
---|
2204 | |
---|
2205 | sys.exit( 1 ) |
---|
2206 | |
---|
2207 | if( DAEMONIZE and USE_SYSLOG ): |
---|
2208 | |
---|
2209 | syslog.openlog( 'jobmond', syslog.LOG_NOWAIT, SYSLOG_FACILITY ) |
---|
2210 | |
---|
2211 | if DAEMONIZE: |
---|
2212 | |
---|
2213 | gather.daemon() |
---|
2214 | else: |
---|
2215 | gather.run() |
---|
2216 | |
---|
2217 | # wh00t? someone started me! :) |
---|
2218 | # |
---|
2219 | if __name__ == '__main__': |
---|
2220 | main() |
---|