Context Navigation

source: trunk/jobarchived/jobarchived.py @ 287

Last change on this file since 287 was 287, checked in by bastiaans, 17 years ago

jobarchived/jobarchived.py:

changed XML retrieval to work with new gmetad
gather XML in same way as webfrontend now
use the same XMLGatherer for both XML processors to prevent unnecessary reconnecting

Property svn:keywords set to Id

File size: 37.8 KB

Rev	Line
[3]	1	#!/usr/bin/env python
[225]	2	#
	3	# This file is part of Jobmonarch
	4	#
	5	# Copyright (C) 2006 Ramon Bastiaans
	6	#
	7	# Jobmonarch is free software; you can redistribute it and/or modify
	8	# it under the terms of the GNU General Public License as published by
	9	# the Free Software Foundation; either version 2 of the License, or
	10	# (at your option) any later version.
	11	#
	12	# Jobmonarch is distributed in the hope that it will be useful,
	13	# but WITHOUT ANY WARRANTY; without even the implied warranty of
	14	# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	15	# GNU General Public License for more details.
	16	#
	17	# You should have received a copy of the GNU General Public License
	18	# along with this program; if not, write to the Free Software
	19	# Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
	20	#
[230]	21	# SVN $Id: jobarchived.py 287 2006-12-21 09:33:19Z bastiaans $
	22	#
[3]	23
[284]	24	DEFAULT_SEARCH_PATH = '/usr/share/jobarchived'
[3]	25
[284]	26	import sys
	27
	28	if DEFAULT_SEARCH_PATH not in sys.path:
	29
	30	sys.path.append( DEFAULT_SEARCH_PATH )
	31
	32	import getopt, syslog, ConfigParser
	33
[214]	34	def processArgs( args ):
[6]	35
[214]	36	SHORT_L = 'c:'
	37	LONG_L = 'config='
[169]	38
[214]	39	config_filename = None
[169]	40
[214]	41	try:
[169]	42
[214]	43	opts, args = getopt.getopt( args, SHORT_L, LONG_L )
[9]	44
[214]	45	except getopt.error, detail:
[60]	46
[214]	47	print detail
	48	sys.exit(1)
[9]	49
[214]	50	for opt, value in opts:
[60]	51
[214]	52	if opt in [ '--config', '-c' ]:
[13]	53
[214]	54	config_filename = value
[198]	55
[214]	56	if not config_filename:
[60]	57
[214]	58	config_filename = '/etc/jobarchived.conf'
[22]	59
[214]	60	try:
	61	return loadConfig( config_filename )
[13]	62
[214]	63	except ConfigParser.NoOptionError, detail:
	64
	65	print detail
	66	sys.exit( 1 )
	67
	68	def loadConfig( filename ):
	69
	70	def getlist( cfg_string ):
	71
	72	my_list = [ ]
	73
	74	for item_txt in cfg_string.split( ',' ):
	75
	76	sep_char = None
	77
	78	item_txt = item_txt.strip()
	79
	80	for s_char in [ "'", '"' ]:
	81
	82	if item_txt.find( s_char ) != -1:
	83
	84	if item_txt.count( s_char ) != 2:
	85
	86	print 'Missing quote: %s' %item_txt
	87	sys.exit( 1 )
	88
	89	else:
	90
	91	sep_char = s_char
	92	break
	93
	94	if sep_char:
	95
	96	item_txt = item_txt.split( sep_char )[1]
	97
	98	my_list.append( item_txt )
	99
	100	return my_list
	101
	102	cfg = ConfigParser.ConfigParser()
	103
	104	cfg.read( filename )
	105
[224]	106	global DEBUG_LEVEL, USE_SYSLOG, SYSLOG_LEVEL, SYSLOG_FACILITY, GMETAD_CONF, ARCHIVE_XMLSOURCE, ARCHIVE_DATASOURCES, ARCHIVE_PATH, ARCHIVE_HOURS_PER_RRD, ARCHIVE_EXCLUDE_METRICS, JOB_SQL_DBASE, DAEMONIZE, RRDTOOL
[214]	107
	108	ARCHIVE_PATH = cfg.get( 'DEFAULT', 'ARCHIVE_PATH' )
	109
	110	ARCHIVE_HOURS_PER_RRD = cfg.getint( 'DEFAULT', 'ARCHIVE_HOURS_PER_RRD' )
	111
	112	DEBUG_LEVEL = cfg.getint( 'DEFAULT', 'DEBUG_LEVEL' )
	113
	114	USE_SYSLOG = cfg.getboolean( 'DEFAULT', 'USE_SYSLOG' )
	115
	116	SYSLOG_LEVEL = cfg.getint( 'DEFAULT', 'SYSLOG_LEVEL' )
	117
	118	try:
	119
	120	SYSLOG_FACILITY = eval( 'syslog.LOG_' + cfg.get( 'DEFAULT', 'SYSLOG_FACILITY' ) )
	121
	122	except AttributeError, detail:
	123
	124	print 'Unknown syslog facility'
	125	sys.exit( 1 )
	126
	127	GMETAD_CONF = cfg.get( 'DEFAULT', 'GMETAD_CONF' )
	128
	129	ARCHIVE_XMLSOURCE = cfg.get( 'DEFAULT', 'ARCHIVE_XMLSOURCE' )
	130
	131	ARCHIVE_DATASOURCES = getlist( cfg.get( 'DEFAULT', 'ARCHIVE_DATASOURCES' ) )
	132
	133	ARCHIVE_EXCLUDE_METRICS = getlist( cfg.get( 'DEFAULT', 'ARCHIVE_EXCLUDE_METRICS' ) )
	134
	135	JOB_SQL_DBASE = cfg.get( 'DEFAULT', 'JOB_SQL_DBASE' )
	136
	137	DAEMONIZE = cfg.getboolean( 'DEFAULT', 'DAEMONIZE' )
	138
[224]	139	RRDTOOL = cfg.get( 'DEFAULT', 'RRDTOOL' )
	140
[214]	141	return True
	142
[17]	143	# What XML data types not to store
[13]	144	#
[17]	145	UNSUPPORTED_ARCHIVE_TYPES = [ 'string' ]
[9]	146
[47]	147	# Maximum time (in seconds) a parsethread may run
	148	#
	149	PARSE_TIMEOUT = 60
	150
	151	# Maximum time (in seconds) a storethread may run
	152	#
	153	STORE_TIMEOUT = 360
	154
[8]	155	"""
[224]	156	The Job Archiving Daemon
[8]	157	"""
	158
[214]	159	from types import *
	160
	161	import DBClass
	162	import xml.sax, xml.sax.handler, socket, string, os, os.path, time, thread, threading, random, re
	163
[84]	164	class DataSQLStore:
	165
	166	db_vars = None
	167	dbc = None
	168
	169	def __init__( self, hostname, database ):
	170
	171	self.db_vars = DBClass.InitVars(DataBaseName=database,
	172	User='root',
	173	Host=hostname,
	174	Password='',
	175	Dictionary='true')
	176
	177	try:
	178	self.dbc = DBClass.DB(self.db_vars)
	179	except DBClass.DBError, details:
[169]	180	debug_msg( 0, 'FATAL ERROR: Unable to connect to database!: ' +str(details) )
[84]	181	sys.exit(1)
	182
	183	def setDatabase(self, statement):
	184	ret = self.doDatabase('set', statement)
	185	return ret
	186
	187	def getDatabase(self, statement):
	188	ret = self.doDatabase('get', statement)
	189	return ret
	190
	191	def doDatabase(self, type, statement):
	192
	193	debug_msg( 6, 'doDatabase(): %s: %s' %(type, statement) )
	194	try:
	195	if type == 'set':
	196	result = self.dbc.Set( statement )
	197	self.dbc.Commit()
	198	elif type == 'get':
	199	result = self.dbc.Get( statement )
	200
	201	except DBClass.DBError, detail:
	202	operation = statement.split(' ')[0]
[169]	203	debug_msg( 0, 'FATAL ERROR: ' +operation+ ' on database failed while doing ['+statement+'] full msg: '+str(detail) )
[84]	204	sys.exit(1)
	205
	206	debug_msg( 6, 'doDatabase(): result: %s' %(result) )
	207	return result
	208
[191]	209	def getJobNodeId( self, job_id, node_id ):
	210
	211	id = self.getDatabase( "SELECT job_id,node_id FROM job_nodes WHERE job_id = '%s' AND node_id = '%s'" %(job_id, node_id) )
	212	if len( id ) > 0:
	213
	214	if len( id[0] ) > 0 and id[0] != '':
	215
	216	return 1
	217
	218	return 0
	219
[84]	220	def getNodeId( self, hostname ):
	221
[89]	222	id = self.getDatabase( "SELECT node_id FROM nodes WHERE node_hostname = '%s'" %hostname )
[84]	223
[89]	224	if len( id ) > 0:
[84]	225
[89]	226	id = id[0][0]
	227
[84]	228	return id
	229	else:
	230	return None
	231
	232	def getNodeIds( self, hostnames ):
	233
	234	ids = [ ]
	235
	236	for node in hostnames:
	237
	238	id = self.getNodeId( node )
	239
	240	if id:
	241	ids.append( id )
	242
	243	return ids
	244
	245	def getJobId( self, jobid ):
	246
	247	id = self.getDatabase( "SELECT job_id FROM jobs WHERE job_id = '%s'" %jobid )
	248
	249	if id:
[89]	250	id = id[0][0]
[84]	251
	252	return id
	253	else:
	254	return None
	255
	256	def addJob( self, job_id, jobattrs ):
	257
	258	if not self.getJobId( job_id ):
	259
	260	self.mutateJob( 'insert', job_id, jobattrs )
	261	else:
	262	self.mutateJob( 'update', job_id, jobattrs )
	263
	264	def mutateJob( self, action, job_id, jobattrs ):
	265
	266	job_values = [ 'name', 'queue', 'owner', 'requested_time', 'requested_memory', 'ppn', 'status', 'start_timestamp', 'stop_timestamp' ]
	267
	268	insert_col_str = 'job_id'
	269	insert_val_str = "'%s'" %job_id
	270	update_str = None
	271
	272	debug_msg( 6, 'mutateJob(): %s %s' %(action,job_id))
	273
[99]	274	ids = [ ]
	275
[84]	276	for valname, value in jobattrs.items():
	277
[96]	278	if valname in job_values and value != '':
[84]	279
	280	column_name = 'job_' + valname
	281
	282	if action == 'insert':
	283
	284	if not insert_col_str:
	285	insert_col_str = column_name
	286	else:
	287	insert_col_str = insert_col_str + ',' + column_name
	288
	289	if not insert_val_str:
	290	insert_val_str = value
	291	else:
	292	insert_val_str = insert_val_str + ",'%s'" %value
	293
	294	elif action == 'update':
	295
	296	if not update_str:
	297	update_str = "%s='%s'" %(column_name, value)
	298	else:
	299	update_str = update_str + ",%s='%s'" %(column_name, value)
	300
[90]	301	elif valname == 'nodes' and value:
[84]	302
[191]	303	node_valid = 1
[190]	304
	305	if len(value) == 1:
	306
[191]	307	if jobattrs['status'] == 'Q':
[190]	308
[191]	309	node_valid = 0
[190]	310
[191]	311	else:
[190]	312
[191]	313	node_valid = 0
[190]	314
[191]	315	for node_char in str(value[0]):
[190]	316
[191]	317	if string.find( string.digits, node_char ) != -1 and not node_valid:
[190]	318
[191]	319	node_valid = 1
[84]	320
[191]	321	if node_valid:
	322
	323	ids = self.addNodes( value, jobattrs['domain'] )
	324
[84]	325	if action == 'insert':
	326
	327	self.setDatabase( "INSERT INTO jobs ( %s ) VALUES ( %s )" %( insert_col_str, insert_val_str ) )
[86]	328
[84]	329	elif action == 'update':
	330
[89]	331	self.setDatabase( "UPDATE jobs SET %s WHERE job_id=%s" %(update_str, job_id) )
[84]	332
[191]	333	if len( ids ) > 0:
	334	self.addJobNodes( job_id, ids )
[190]	335
[154]	336	def addNodes( self, hostnames, domain ):
[84]	337
[98]	338	ids = [ ]
	339
[84]	340	for node in hostnames:
	341
[154]	342	node = '%s.%s' %( node, domain )
[84]	343	id = self.getNodeId( node )
	344
	345	if not id:
	346	self.setDatabase( "INSERT INTO nodes ( node_hostname ) VALUES ( '%s' )" %node )
[98]	347	id = self.getNodeId( node )
[84]	348
[98]	349	ids.append( id )
	350
	351	return ids
	352
[86]	353	def addJobNodes( self, jobid, nodes ):
	354
	355	for node in nodes:
	356
[191]	357	if not self.getJobNodeId( jobid, node ):
	358
	359	self.addJobNode( jobid, node )
	360
[84]	361	def addJobNode( self, jobid, nodeid ):
	362
	363	self.setDatabase( "INSERT INTO job_nodes (job_id,node_id) VALUES ( %s,%s )" %(jobid, nodeid) )
	364
	365	def storeJobInfo( self, jobid, jobattrs ):
	366
	367	self.addJob( jobid, jobattrs )
	368
[37]	369	class RRDMutator:
[63]	370	"""A class for performing RRD mutations"""
[37]	371
[277]	372	binary = None
[37]	373
	374	def __init__( self, binary=None ):
[63]	375	"""Set alternate binary if supplied"""
[37]	376
	377	if binary:
	378	self.binary = binary
	379
	380	def create( self, filename, args ):
[63]	381	"""Create a new rrd with args"""
	382
[40]	383	return self.perform( 'create', '"' + filename + '"', args )
[37]	384
	385	def update( self, filename, args ):
[63]	386	"""Update a rrd with args"""
	387
[40]	388	return self.perform( 'update', '"' + filename + '"', args )
[37]	389
[42]	390	def grabLastUpdate( self, filename ):
[63]	391	"""Determine the last update time of filename rrd"""
[42]	392
	393	last_update = 0
	394
[53]	395	debug_msg( 8, self.binary + ' info "' + filename + '"' )
	396
[42]	397	for line in os.popen( self.binary + ' info "' + filename + '"' ).readlines():
	398
	399	if line.find( 'last_update') != -1:
	400
	401	last_update = line.split( ' = ' )[1]
	402
	403	if last_update:
	404	return last_update
	405	else:
	406	return 0
	407
[40]	408	def perform( self, action, filename, args ):
[63]	409	"""Perform action on rrd filename with args"""
[37]	410
	411	arg_string = None
	412
[40]	413	if type( args ) is not ListType:
	414	debug_msg( 8, 'Arguments needs to be of type List' )
	415	return 1
	416
[37]	417	for arg in args:
	418
	419	if not arg_string:
	420
	421	arg_string = arg
	422	else:
	423	arg_string = arg_string + ' ' + arg
	424
[54]	425	debug_msg( 8, self.binary + ' ' + action + ' ' + filename + ' ' + arg_string )
[37]	426
[146]	427	cmd = os.popen( self.binary + ' ' + action + ' ' + filename + ' ' + arg_string )
	428	lines = cmd.readlines()
	429	cmd.close()
[37]	430
[146]	431	for line in lines:
	432
[37]	433	if line.find( 'ERROR' ) != -1:
	434
	435	error_msg = string.join( line.split( ' ' )[1:] )
[40]	436	debug_msg( 8, error_msg )
[37]	437	return 1
	438
	439	return 0
	440
[78]	441	class XMLProcessor:
	442	"""Skeleton class for XML processor's"""
	443
	444	def run( self ):
	445	"""Do main processing of XML here"""
	446
	447	pass
	448
	449	class TorqueXMLProcessor( XMLProcessor ):
	450	"""Main class for processing XML and acting with it"""
	451
[287]	452	def __init__( self, XMLSource ):
[78]	453	"""Setup initial XML connection and handlers"""
	454
[287]	455	self.myXMLGatherer = XMLGatherer( ARCHIVE_XMLSOURCE.split( ':' )[0], ARCHIVE_XMLSOURCE.split( ':' )[1] )
	456	#self.myXMLSource = self.myXMLGatherer.getFileObject()
	457	self.myXMLSource = XMLSource
	458	print self.myXMLSource
	459	self.myXMLHandler = TorqueXMLHandler()
	460	self.myXMLError = XMLErrorHandler()
[78]	461
[287]	462	self.config = GangliaConfigParser( GMETAD_CONF )
	463
[78]	464	def run( self ):
	465	"""Main XML processing"""
	466
[169]	467	debug_msg( 1, 'torque_xml_thread(): started.' )
[87]	468
[78]	469	while( 1 ):
	470
[287]	471	#self.myXMLSource = self.mXMLGatherer.getFileObject()
[169]	472	debug_msg( 1, 'torque_xml_thread(): Parsing..' )
[176]	473
[287]	474	my_data = self.myXMLSource.getData()
	475
[176]	476	try:
[287]	477	xml.sax.parseString( my_data, self.myXMLHandler, self.myXMLError )
[176]	478	except socket.error, msg:
	479	debug_msg( 0, 'ERROR: Socket error in connect to datasource!: %s' %msg )
	480
[169]	481	debug_msg( 1, 'torque_xml_thread(): Done parsing.' )
	482	debug_msg( 1, 'torque_xml_thread(): Sleeping.. (%ss)' %(str( self.config.getLowestInterval() ) ) )
[87]	483	time.sleep( self.config.getLowestInterval() )
[78]	484
[71]	485	class TorqueXMLHandler( xml.sax.handler.ContentHandler ):
[63]	486	"""Parse Torque's jobinfo XML from our plugin"""
	487
[72]	488	jobAttrs = { }
	489
[84]	490	def __init__( self ):
	491
[214]	492	self.ds = DataSQLStore( JOB_SQL_DBASE.split( '/' )[0], JOB_SQL_DBASE.split( '/' )[1] )
[98]	493	self.jobs_processed = [ ]
	494	self.jobs_to_store = [ ]
[84]	495
[183]	496	def startDocument( self ):
	497
	498	self.heartbeat = 0
	499
[63]	500	def startElement( self, name, attrs ):
	501	"""
	502	This XML will be all gmetric XML
	503	so there will be no specific start/end element
	504	just one XML statement with all info
	505	"""
	506
[73]	507	jobinfo = { }
[63]	508
[199]	509	if name == 'CLUSTER':
[63]	510
[199]	511	self.clustername = attrs.get( 'NAME', "" )
	512
	513	elif name == 'METRIC' and self.clustername in ARCHIVE_DATASOURCES:
	514
[70]	515	metricname = attrs.get( 'NAME', "" )
[63]	516
[285]	517	if metricname == 'MONARCH-HEARTBEAT':
[70]	518	self.heartbeat = attrs.get( 'VAL', "" )
[63]	519
[285]	520	elif metricname.find( 'MONARCH-JOB' ) != -1:
[63]	521
[286]	522	job_id = metricname.split( 'MONARCH-JOB-' )[1].split( '-' )[0]
[63]	523	val = attrs.get( 'VAL', "" )
	524
[96]	525	if not job_id in self.jobs_processed:
	526	self.jobs_processed.append( job_id )
	527
[73]	528	check_change = 0
	529
	530	if self.jobAttrs.has_key( job_id ):
	531	check_change = 1
	532
[63]	533	valinfo = val.split( ' ' )
	534
	535	for myval in valinfo:
	536
[84]	537	if len( myval.split( '=' ) ) > 1:
[63]	538
[84]	539	valname = myval.split( '=' )[0]
	540	value = myval.split( '=' )[1]
[70]	541
[84]	542	if valname == 'nodes':
	543	value = value.split( ';' )
[72]	544
[84]	545	jobinfo[ valname ] = value
	546
[73]	547	if check_change:
[182]	548	if self.jobinfoChanged( self.jobAttrs, job_id, jobinfo ) and self.jobAttrs[ job_id ]['status'] in [ 'R', 'Q' ]:
[100]	549	self.jobAttrs[ job_id ]['stop_timestamp'] = ''
[82]	550	self.jobAttrs[ job_id ] = self.setJobAttrs( self.jobAttrs[ job_id ], jobinfo )
[84]	551	if not job_id in self.jobs_to_store:
	552	self.jobs_to_store.append( job_id )
	553
[102]	554	debug_msg( 6, 'jobinfo for job %s has changed' %job_id )
[73]	555	else:
	556	self.jobAttrs[ job_id ] = jobinfo
[84]	557
	558	if not job_id in self.jobs_to_store:
	559	self.jobs_to_store.append( job_id )
	560
[102]	561	debug_msg( 6, 'jobinfo for job %s has changed' %job_id )
[73]	562
[77]	563	def endDocument( self ):
[74]	564	"""When all metrics have gone, check if any jobs have finished"""
[72]	565
[182]	566	if self.heartbeat:
	567	for jobid, jobinfo in self.jobAttrs.items():
[74]	568
[182]	569	# This is an old job, not in current jobinfo list anymore
	570	# it must have finished, since we _did_ get a new heartbeat
	571	#
	572	mytime = int( jobinfo['reported'] ) + int( jobinfo['poll_interval'] )
[102]	573
[184]	574	if (mytime < self.heartbeat) and (jobid not in self.jobs_processed) and (jobinfo['status'] == 'R'):
[74]	575
[182]	576	if not jobid in self.jobs_processed:
	577	self.jobs_processed.append( jobid )
[96]	578
[182]	579	self.jobAttrs[ jobid ]['status'] = 'F'
	580	self.jobAttrs[ jobid ]['stop_timestamp'] = str( mytime )
[96]	581
[182]	582	if not jobid in self.jobs_to_store:
	583	self.jobs_to_store.append( jobid )
[74]	584
[182]	585	debug_msg( 1, 'torque_xml_thread(): Storing..' )
[87]	586
[182]	587	for jobid in self.jobs_to_store:
	588	if self.jobAttrs[ jobid ]['status'] in [ 'R', 'Q', 'F' ]:
[84]	589
[184]	590	self.ds.storeJobInfo( jobid, self.jobAttrs[ jobid ] )
	591
	592	if self.jobAttrs[ jobid ]['status'] == 'F':
	593	del self.jobAttrs[ jobid ]
	594
[182]	595	debug_msg( 1, 'torque_xml_thread(): Done storing.' )
[87]	596
[182]	597	self.jobs_processed = [ ]
	598	self.jobs_to_store = [ ]
[84]	599
[82]	600	def setJobAttrs( self, old, new ):
	601	"""
	602	Set new job attributes in old, but not lose existing fields
	603	if old attributes doesn't have those
	604	"""
	605
	606	for valname, value in new.items():
	607	old[ valname ] = value
	608
	609	return old
	610
	611
[73]	612	def jobinfoChanged( self, jobattrs, jobid, jobinfo ):
[74]	613	"""
	614	Check if jobinfo has changed from jobattrs[jobid]
	615	if it's report time is bigger than previous one
	616	and it is report time is recent (equal to heartbeat)
	617	"""
[72]	618
[87]	619	ignore_changes = [ 'reported' ]
	620
[73]	621	if jobattrs.has_key( jobid ):
	622
	623	for valname, value in jobinfo.items():
	624
[87]	625	if valname not in ignore_changes:
[73]	626
[87]	627	if jobattrs[ jobid ].has_key( valname ):
[73]	628
[87]	629	if value != jobattrs[ jobid ][ valname ]:
[73]	630
[87]	631	if jobinfo['reported'] > jobattrs[ jobid ][ 'reported' ] and jobinfo['reported'] == self.heartbeat:
	632	return 1
[73]	633
[87]	634	else:
	635	return 1
	636
[73]	637	return 0
	638
[71]	639	class GangliaXMLHandler( xml.sax.handler.ContentHandler ):
[63]	640	"""Parse Ganglia's XML"""
[3]	641
[33]	642	def __init__( self, config ):
[63]	643	"""Setup initial variables and gather info on existing rrd archive"""
	644
[33]	645	self.config = config
[35]	646	self.clusters = { }
[169]	647	debug_msg( 1, 'Checking existing toga rrd archive..' )
[44]	648	self.gatherClusters()
[169]	649	debug_msg( 1, 'Check done.' )
[33]	650
[44]	651	def gatherClusters( self ):
[63]	652	"""Find all existing clusters in archive dir"""
[44]	653
[45]	654	archive_dir = check_dir(ARCHIVE_PATH)
[44]	655
	656	hosts = [ ]
	657
	658	if os.path.exists( archive_dir ):
	659
	660	dirlist = os.listdir( archive_dir )
	661
	662	for item in dirlist:
	663
	664	clustername = item
	665
[60]	666	if not self.clusters.has_key( clustername ) and clustername in ARCHIVE_DATASOURCES:
[44]	667
	668	self.clusters[ clustername ] = RRDHandler( self.config, clustername )
	669
[6]	670	def startElement( self, name, attrs ):
[63]	671	"""Memorize appropriate data from xml start tags"""
[3]	672
[7]	673	if name == 'GANGLIA_XML':
[32]	674
	675	self.XMLSource = attrs.get( 'SOURCE', "" )
	676	self.gangliaVersion = attrs.get( 'VERSION', "" )
	677
[12]	678	debug_msg( 10, 'Found XML data: source %s version %s' %( self.XMLSource, self.gangliaVersion ) )
[6]	679
[7]	680	elif name == 'GRID':
[32]	681
	682	self.gridName = attrs.get( 'NAME', "" )
	683	self.time = attrs.get( 'LOCALTIME', "" )
	684
[12]	685	debug_msg( 10, '`-Grid found: %s' %( self.gridName ) )
[6]	686
[7]	687	elif name == 'CLUSTER':
[32]	688
	689	self.clusterName = attrs.get( 'NAME', "" )
	690	self.time = attrs.get( 'LOCALTIME', "" )
	691
[60]	692	if not self.clusters.has_key( self.clusterName ) and self.clusterName in ARCHIVE_DATASOURCES:
[32]	693
[34]	694	self.clusters[ self.clusterName ] = RRDHandler( self.config, self.clusterName )
[33]	695
[35]	696	debug_msg( 10, ' \|-Cluster found: %s' %( self.clusterName ) )
[6]	697
[60]	698	elif name == 'HOST' and self.clusterName in ARCHIVE_DATASOURCES:
[32]	699
	700	self.hostName = attrs.get( 'NAME', "" )
	701	self.hostIp = attrs.get( 'IP', "" )
	702	self.hostReported = attrs.get( 'REPORTED', "" )
	703
[12]	704	debug_msg( 10, ' \| \|-Host found: %s - ip %s reported %s' %( self.hostName, self.hostIp, self.hostReported ) )
[6]	705
[60]	706	elif name == 'METRIC' and self.clusterName in ARCHIVE_DATASOURCES:
[6]	707
[32]	708	type = attrs.get( 'TYPE', "" )
[198]	709
	710	exclude_metric = False
	711
	712	for ex_metricstr in ARCHIVE_EXCLUDE_METRICS:
[6]	713
[198]	714	orig_name = attrs.get( 'NAME', "" )
[3]	715
[198]	716	if string.lower( orig_name ) == string.lower( ex_metricstr ):
	717
	718	exclude_metric = True
	719
	720	elif re.match( ex_metricstr, orig_name ):
	721
	722	exclude_metric = True
	723
	724	if type not in UNSUPPORTED_ARCHIVE_TYPES and not exclude_metric:
	725
[32]	726	myMetric = { }
	727	myMetric['name'] = attrs.get( 'NAME', "" )
	728	myMetric['val'] = attrs.get( 'VAL', "" )
	729	myMetric['time'] = self.hostReported
[3]	730
[34]	731	self.clusters[ self.clusterName ].memMetric( self.hostName, myMetric )
[3]	732
[34]	733	debug_msg( 11, ' \| \| \|-metric: %s:%s' %( myMetric['name'], myMetric['val'] ) )
[6]	734
[34]	735	def storeMetrics( self ):
[63]	736	"""Store metrics of each cluster rrd handler"""
[9]	737
[34]	738	for clustername, rrdh in self.clusters.items():
[16]	739
[38]	740	ret = rrdh.storeMetrics()
[9]	741
[38]	742	if ret:
	743	debug_msg( 9, 'An error occured while storing metrics for cluster %s' %clustername )
	744	return 1
	745
	746	return 0
	747
[71]	748	class XMLErrorHandler( xml.sax.handler.ErrorHandler ):
	749
	750	def error( self, exception ):
	751	"""Recoverable error"""
	752
[169]	753	debug_msg( 0, 'Recoverable XML error ' + str( exception ) + ' ignored.' )
[71]	754
	755	def fatalError( self, exception ):
	756	"""Non-recoverable error"""
	757
	758	exception_str = str( exception )
	759
	760	# Ignore 'no element found' errors
	761	if exception_str.find( 'no element found' ) != -1:
[169]	762	debug_msg( 0, 'No XML data found: Socket not (re)connected or datasource not available.' )
[71]	763	return 0
	764
[170]	765	debug_msg( 0, 'FATAL ERROR: Non-recoverable XML error ' + str( exception ) )
[71]	766	sys.exit( 1 )
	767
	768	def warning( self, exception ):
	769	"""Warning"""
	770
	771	debug_msg( 0, 'Warning ' + str( exception ) )
	772
[78]	773	class XMLGatherer:
[63]	774	"""Setup a connection and file object to Ganglia's XML"""
[3]	775
[287]	776	s = None
	777	fd = None
	778	data = None
[8]	779
[287]	780	# Time since the last update
	781	#
	782	LAST_UPDATE = 0
	783
	784	# Minimum interval between updates
	785	#
	786	MIN_UPDATE_INT = 10
	787
	788	# Is a update occuring now
	789	#
	790	update_now = False
	791
[8]	792	def __init__( self, host, port ):
[63]	793	"""Store host and port for connection"""
[8]	794
[5]	795	self.host = host
	796	self.port = port
[3]	797
[287]	798	self.retrieveData()
	799
	800	def retrieveData( self ):
[63]	801	"""Setup connection to XML source"""
[8]	802
[287]	803	self.update_now = True
	804
[8]	805	for res in socket.getaddrinfo( self.host, self.port, socket.AF_UNSPEC, socket.SOCK_STREAM ):
[32]	806
[5]	807	af, socktype, proto, canonname, sa = res
[32]	808
[5]	809	try:
[32]	810
[8]	811	self.s = socket.socket( af, socktype, proto )
[32]	812
[5]	813	except socket.error, msg:
[32]	814
[8]	815	self.s = None
[5]	816	continue
[32]	817
[287]	818	print self.s
	819
[5]	820	try:
[32]	821
[8]	822	self.s.connect( sa )
[32]	823
[5]	824	except socket.error, msg:
[32]	825
[70]	826	self.disconnect()
[5]	827	continue
[32]	828
[5]	829	break
[3]	830
[287]	831	print self.s
	832
[8]	833	if self.s is None:
[32]	834
[170]	835	debug_msg( 0, 'FATAL ERROR: Could not open socket or unable to connect to datasource!' )
[287]	836	self.update_now = False
[33]	837	sys.exit( 1 )
[5]	838
[287]	839	else:
	840	self.s.send( '\n' )
	841
	842	my_fp = self.s.makefile( 'r' )
	843	my_data = my_fp.readlines()
	844	my_data = string.join( my_data, '' )
	845
	846	self.data = my_data
	847
	848	self.LAST_UPDATE = time.time()
	849
	850	self.update_now = False
	851
[33]	852	def disconnect( self ):
[63]	853	"""Close socket"""
[33]	854
	855	if self.s:
[287]	856	#self.s.shutdown( 2 )
[33]	857	self.s.close()
	858	self.s = None
	859
	860	def __del__( self ):
[63]	861	"""Kill the socket before we leave"""
[33]	862
	863	self.disconnect()
	864
[287]	865	def reGetData( self ):
[70]	866	"""Reconnect"""
[33]	867
[287]	868	while self.update_now:
	869
	870	# Must be another update in progress:
	871	# Wait until the update is complete
	872	#
	873	time.sleep( 1 )
	874
[38]	875	if self.s:
	876	self.disconnect()
[33]	877
[287]	878	self.retrieveData()
[5]	879
[287]	880	def getData( self ):
	881
	882	"""Return the XML data"""
	883
	884	# If more than MIN_UPDATE_INT seconds passed since last data update
	885	# update the XML first before returning it
	886	#
	887
	888	cur_time = time.time()
	889
	890	if ( cur_time - self.LAST_UPDATE ) > self.MIN_UPDATE_INT:
	891
	892	self.reGetData()
	893
	894	while self.update_now:
	895
	896	# Must be another update in progress:
	897	# Wait until the update is complete
	898	#
	899	time.sleep( 1 )
	900
	901	return self.data
	902
[70]	903	def makeFileDescriptor( self ):
	904	"""Make file descriptor that points to our socket connection"""
	905
	906	self.reconnect()
	907
	908	if self.s:
	909	self.fd = self.s.makefile( 'r' )
	910
	911	def getFileObject( self ):
	912	"""Connect, and return a file object"""
	913
[78]	914	self.makeFileDescriptor()
	915
[70]	916	if self.fd:
	917	return self.fd
	918
[78]	919	class GangliaXMLProcessor( XMLProcessor ):
[63]	920	"""Main class for processing XML and acting with it"""
[5]	921
[287]	922	def __init__( self, XMLSource ):
[63]	923	"""Setup initial XML connection and handlers"""
[33]	924
[287]	925	self.config = GangliaConfigParser( GMETAD_CONF )
[33]	926
[287]	927	self.myXMLGatherer = XMLGatherer( ARCHIVE_XMLSOURCE.split( ':' )[0], ARCHIVE_XMLSOURCE.split( ':' )[1] )
	928	#self.myXMLSource = self.myXMLGatherer.getFileObject()
	929	self.myXMLSource = XMLSource
	930	print self.myXMLSource
	931	self.myXMLHandler = GangliaXMLHandler( self.config )
	932	self.myXMLError = XMLErrorHandler()
[73]	933
[9]	934	def run( self ):
[63]	935	"""Main XML processing; start a xml and storethread"""
[8]	936
[102]	937	xml_thread = threading.Thread( None, self.processXML, 'xmlthread' )
	938	store_thread = threading.Thread( None, self.storeMetrics, 'storethread' )
[22]	939
[36]	940	while( 1 ):
	941
[102]	942	if not xml_thread.isAlive():
[36]	943	# Gather XML at the same interval as gmetad
	944
	945	# threaded call to: self.processXML()
	946	#
[169]	947	try:
	948	xml_thread = threading.Thread( None, self.processXML, 'xml_thread' )
	949	xml_thread.start()
[176]	950	except thread.error, msg:
[169]	951	debug_msg( 0, 'ERROR: Unable to start xml_thread!: '+str(msg))
	952	#return 1
[36]	953
[102]	954	if not store_thread.isAlive():
[55]	955	# Store metrics every .. sec
[36]	956
[55]	957	# threaded call to: self.storeMetrics()
	958	#
[169]	959	try:
	960	store_thread = threading.Thread( None, self.storeMetrics, 'store_thread' )
	961	store_thread.start()
[176]	962	except thread.error, msg:
[169]	963	debug_msg( 0, 'ERROR: Unable to start store_thread!: '+str(msg))
	964	#return 1
[36]	965
	966	# Just sleep a sec here, to prevent daemon from going mad. We're all threads here anyway
	967	time.sleep( 1 )
	968
[33]	969	def storeMetrics( self ):
[63]	970	"""Store metrics retained in memory to disk"""
[22]	971
[63]	972	# Store metrics somewhere between every 360 and 640 seconds
[38]	973	#
[55]	974	STORE_INTERVAL = random.randint( 360, 640 )
[22]	975
[169]	976	try:
	977	store_metric_thread = threading.Thread( None, self.storeThread, 'store_metric_thread' )
	978	store_metric_thread.start()
[176]	979	except thread.error, msg:
[169]	980	debug_msg( 0, 'ERROR: Unable to start ganglia_store_thread()!: '+str(msg) )
	981	return 1
[36]	982
[169]	983	debug_msg( 1, 'ganglia_store_thread(): started.' )
	984
	985	debug_msg( 1, 'ganglia_store_thread(): Sleeping.. (%ss)' %STORE_INTERVAL )
[36]	986	time.sleep( STORE_INTERVAL )
[169]	987	debug_msg( 1, 'ganglia_store_thread(): Done sleeping.' )
[36]	988
[102]	989	if store_metric_thread.isAlive():
[36]	990
[169]	991	debug_msg( 1, 'ganglia_store_thread(): storemetricthread() still running, waiting to finish..' )
[136]	992	store_metric_thread.join( STORE_TIMEOUT ) # Maximum time is for storing thread to finish
[169]	993	debug_msg( 1, 'ganglia_store_thread(): Done waiting.' )
[36]	994
[169]	995	debug_msg( 1, 'ganglia_store_thread(): finished.' )
[36]	996
	997	return 0
	998
[39]	999	def storeThread( self ):
[63]	1000	"""Actual metric storing thread"""
[39]	1001
[169]	1002	debug_msg( 1, 'ganglia_store_metric_thread(): started.' )
	1003	debug_msg( 1, 'ganglia_store_metric_thread(): Storing data..' )
[78]	1004	ret = self.myXMLHandler.storeMetrics()
[176]	1005	if ret > 0:
	1006	debug_msg( 0, 'ganglia_store_metric_thread(): UNKNOWN ERROR %s while storing Metrics!' %str(ret) )
[169]	1007	debug_msg( 1, 'ganglia_store_metric_thread(): Done storing.' )
	1008	debug_msg( 1, 'ganglia_store_metric_thread(): finished.' )
[39]	1009
[176]	1010	return 0
[39]	1011
[8]	1012	def processXML( self ):
[63]	1013	"""Process XML"""
[8]	1014
[169]	1015	try:
	1016	parsethread = threading.Thread( None, self.parseThread, 'parsethread' )
	1017	parsethread.start()
[176]	1018	except thread.error, msg:
[169]	1019	debug_msg( 0, 'ERROR: Unable to start ganglia_xml_thread()!: ' + str(msg) )
	1020	return 1
[8]	1021
[169]	1022	debug_msg( 1, 'ganglia_xml_thread(): started.' )
[36]	1023
[169]	1024	debug_msg( 1, 'ganglia_xml_thread(): Sleeping.. (%ss)' %self.config.getLowestInterval() )
[36]	1025	time.sleep( float( self.config.getLowestInterval() ) )
[169]	1026	debug_msg( 1, 'ganglia_xml_thread(): Done sleeping.' )
[36]	1027
	1028	if parsethread.isAlive():
	1029
[169]	1030	debug_msg( 1, 'ganglia_xml_thread(): parsethread() still running, waiting (%ss) to finish..' %PARSE_TIMEOUT )
[47]	1031	parsethread.join( PARSE_TIMEOUT ) # Maximum time for XML thread to finish
[169]	1032	debug_msg( 1, 'ganglia_xml_thread(): Done waiting.' )
[36]	1033
[169]	1034	debug_msg( 1, 'ganglia_xml_thread(): finished.' )
[36]	1035
	1036	return 0
	1037
[39]	1038	def parseThread( self ):
[63]	1039	"""Actual parsing thread"""
[39]	1040
[169]	1041	debug_msg( 1, 'ganglia_parse_thread(): started.' )
	1042	debug_msg( 1, 'ganglia_parse_thread(): Parsing XML..' )
[287]	1043	#self.myXMLSource = self.myXMLGatherer.getFileObject()
	1044
	1045	my_data = self.myXMLSource.getData()
[176]	1046
	1047	try:
[287]	1048	xml.sax.parseString( my_data, self.myXMLHandler, self.myXMLError )
[176]	1049	except socket.error, msg:
	1050	debug_msg( 0, 'ERROR: Socket error in connect to datasource!: %s' %msg )
	1051
[169]	1052	debug_msg( 1, 'ganglia_parse_thread(): Done parsing.' )
	1053	debug_msg( 1, 'ganglia_parse_thread(): finished.' )
[39]	1054
[176]	1055	return 0
[39]	1056
[9]	1057	class GangliaConfigParser:
	1058
[34]	1059	sources = [ ]
[9]	1060
	1061	def __init__( self, config ):
[63]	1062	"""Parse some stuff from our gmetad's config, such as polling interval"""
[32]	1063
[9]	1064	self.config = config
	1065	self.parseValues()
	1066
[32]	1067	def parseValues( self ):
[63]	1068	"""Parse certain values from gmetad.conf"""
[9]	1069
	1070	readcfg = open( self.config, 'r' )
	1071
	1072	for line in readcfg.readlines():
	1073
	1074	if line.count( '"' ) > 1:
	1075
[10]	1076	if line.find( 'data_source' ) != -1 and line[0] != '#':
[9]	1077
[11]	1078	source = { }
	1079	source['name'] = line.split( '"' )[1]
[9]	1080	source_words = line.split( '"' )[2].split( ' ' )
	1081
	1082	for word in source_words:
	1083
	1084	valid_interval = 1
	1085
	1086	for letter in word:
[32]	1087
[9]	1088	if letter not in string.digits:
[32]	1089
[9]	1090	valid_interval = 0
	1091
[10]	1092	if valid_interval and len(word) > 0:
[32]	1093
[9]	1094	source['interval'] = word
[12]	1095	debug_msg( 9, 'polling interval for %s = %s' %(source['name'], source['interval'] ) )
[33]	1096
	1097	# No interval found, use Ganglia's default
	1098	if not source.has_key( 'interval' ):
	1099	source['interval'] = 15
	1100	debug_msg( 9, 'polling interval for %s defaulted to 15' %(source['name']) )
[32]	1101
[33]	1102	self.sources.append( source )
[9]	1103
	1104	def getInterval( self, source_name ):
[63]	1105	"""Return interval for source_name"""
[32]	1106
[9]	1107	for source in self.sources:
[32]	1108
[12]	1109	if source['name'] == source_name:
[32]	1110
[9]	1111	return source['interval']
[32]	1112
[9]	1113	return None
	1114
[34]	1115	def getLowestInterval( self ):
[63]	1116	"""Return the lowest interval of all clusters"""
[34]	1117
	1118	lowest_interval = 0
	1119
	1120	for source in self.sources:
	1121
	1122	if not lowest_interval or source['interval'] <= lowest_interval:
	1123
	1124	lowest_interval = source['interval']
	1125
	1126	# Return 15 when nothing is found, so that the daemon won't go insane with 0 sec delays
	1127	if lowest_interval:
	1128	return lowest_interval
	1129	else:
	1130	return 15
	1131
[9]	1132	class RRDHandler:
[63]	1133	"""Class for handling RRD activity"""
[9]	1134
[32]	1135	myMetrics = { }
[40]	1136	lastStored = { }
[47]	1137	timeserials = { }
[36]	1138	slot = None
[32]	1139
[33]	1140	def __init__( self, config, cluster ):
[63]	1141	"""Setup initial variables"""
[78]	1142
[44]	1143	self.block = 0
[9]	1144	self.cluster = cluster
[33]	1145	self.config = config
[40]	1146	self.slot = threading.Lock()
[277]	1147	self.rrdm = RRDMutator( RRDTOOL )
[42]	1148	self.gatherLastUpdates()
[9]	1149
[42]	1150	def gatherLastUpdates( self ):
[63]	1151	"""Populate the lastStored list, containing timestamps of all last updates"""
[42]	1152
	1153	cluster_dir = '%s/%s' %( check_dir(ARCHIVE_PATH), self.cluster )
	1154
	1155	hosts = [ ]
	1156
	1157	if os.path.exists( cluster_dir ):
	1158
[44]	1159	dirlist = os.listdir( cluster_dir )
[42]	1160
[44]	1161	for dir in dirlist:
[42]	1162
[44]	1163	hosts.append( dir )
[42]	1164
	1165	for host in hosts:
	1166
[47]	1167	host_dir = cluster_dir + '/' + host
	1168	dirlist = os.listdir( host_dir )
	1169
	1170	for dir in dirlist:
	1171
	1172	if not self.timeserials.has_key( host ):
	1173
	1174	self.timeserials[ host ] = [ ]
	1175
	1176	self.timeserials[ host ].append( dir )
	1177
[42]	1178	last_serial = self.getLastRrdTimeSerial( host )
	1179	if last_serial:
	1180
	1181	metric_dir = cluster_dir + '/' + host + '/' + last_serial
	1182	if os.path.exists( metric_dir ):
	1183
[44]	1184	dirlist = os.listdir( metric_dir )
[42]	1185
[44]	1186	for file in dirlist:
[42]	1187
[44]	1188	metricname = file.split( '.rrd' )[0]
[42]	1189
[44]	1190	if not self.lastStored.has_key( host ):
[42]	1191
[44]	1192	self.lastStored[ host ] = { }
[42]	1193
[44]	1194	self.lastStored[ host ][ metricname ] = self.rrdm.grabLastUpdate( metric_dir + '/' + file )
[42]	1195
[32]	1196	def getClusterName( self ):
[63]	1197	"""Return clustername"""
	1198
[32]	1199	return self.cluster
	1200
	1201	def memMetric( self, host, metric ):
[63]	1202	"""Store metric from host in memory"""
[32]	1203
[179]	1204	# <ATOMIC>
	1205	#
	1206	self.slot.acquire()
	1207
[34]	1208	if self.myMetrics.has_key( host ):
[32]	1209
[34]	1210	if self.myMetrics[ host ].has_key( metric['name'] ):
[32]	1211
[34]	1212	for mymetric in self.myMetrics[ host ][ metric['name'] ]:
[32]	1213
[34]	1214	if mymetric['time'] == metric['time']:
[32]	1215
[34]	1216	# Allready have this metric, abort
[179]	1217	self.slot.release()
[34]	1218	return 1
	1219	else:
	1220	self.myMetrics[ host ][ metric['name'] ] = [ ]
	1221	else:
[32]	1222	self.myMetrics[ host ] = { }
[34]	1223	self.myMetrics[ host ][ metric['name'] ] = [ ]
[32]	1224
[63]	1225	# Push new metric onto stack
	1226	# atomic code; only 1 thread at a time may access the stack
	1227
[32]	1228	self.myMetrics[ host ][ metric['name'] ].append( metric )
	1229
[40]	1230	self.slot.release()
[53]	1231	#
	1232	# </ATOMIC>
[40]	1233
[47]	1234	def makeUpdateList( self, host, metriclist ):
[63]	1235	"""
	1236	Make a list of update values for rrdupdate
	1237	but only those that we didn't store before
	1238	"""
[37]	1239
	1240	update_list = [ ]
[41]	1241	metric = None
[37]	1242
[47]	1243	while len( metriclist ) > 0:
[37]	1244
[53]	1245	metric = metriclist.pop( 0 )
[37]	1246
[53]	1247	if self.checkStoreMetric( host, metric ):
	1248	update_list.append( '%s:%s' %( metric['time'], metric['val'] ) )
[40]	1249
[37]	1250	return update_list
	1251
[49]	1252	def checkStoreMetric( self, host, metric ):
[63]	1253	"""Check if supplied metric if newer than last one stored"""
[40]	1254
	1255	if self.lastStored.has_key( host ):
	1256
[47]	1257	if self.lastStored[ host ].has_key( metric['name'] ):
[40]	1258
[47]	1259	if metric['time'] <= self.lastStored[ host ][ metric['name'] ]:
[40]	1260
[50]	1261	# This is old
[40]	1262	return 0
	1263
[50]	1264	return 1
	1265
[54]	1266	def memLastUpdate( self, host, metricname, metriclist ):
[63]	1267	"""
	1268	Memorize the time of the latest metric from metriclist
	1269	but only if it wasn't allready memorized
	1270	"""
[50]	1271
[54]	1272	if not self.lastStored.has_key( host ):
	1273	self.lastStored[ host ] = { }
	1274
[50]	1275	last_update_time = 0
	1276
	1277	for metric in metriclist:
	1278
[54]	1279	if metric['name'] == metricname:
[50]	1280
[54]	1281	if metric['time'] > last_update_time:
[50]	1282
[54]	1283	last_update_time = metric['time']
[40]	1284
[54]	1285	if self.lastStored[ host ].has_key( metricname ):
[52]	1286
[54]	1287	if last_update_time <= self.lastStored[ host ][ metricname ]:
[52]	1288	return 1
[40]	1289
[54]	1290	self.lastStored[ host ][ metricname ] = last_update_time
[52]	1291
[33]	1292	def storeMetrics( self ):
[63]	1293	"""
	1294	Store all metrics from memory to disk
	1295	and do it to the RRD's in appropriate timeperiod directory
	1296	"""
[33]	1297
	1298	for hostname, mymetrics in self.myMetrics.items():
	1299
	1300	for metricname, mymetric in mymetrics.items():
	1301
[53]	1302	metrics_to_store = [ ]
	1303
[63]	1304	# Pop metrics from stack for storing until none is left
	1305	# atomic code: only 1 thread at a time may access myMetrics
	1306
[53]	1307	# <ATOMIC>
[50]	1308	#
[47]	1309	self.slot.acquire()
[33]	1310
[54]	1311	while len( self.myMetrics[ hostname ][ metricname ] ) > 0:
[53]	1312
[54]	1313	if len( self.myMetrics[ hostname ][ metricname ] ) > 0:
[53]	1314
[176]	1315	try:
	1316	metrics_to_store.append( self.myMetrics[ hostname ][ metricname ].pop( 0 ) )
	1317	except IndexError, msg:
	1318
[179]	1319	# Somehow sometimes myMetrics[ hostname ][ metricname ]
	1320	# is still len 0 when the statement is executed.
	1321	# Just ignore indexerror's..
[176]	1322	pass
	1323
[53]	1324	self.slot.release()
	1325	#
	1326	# </ATOMIC>
	1327
[47]	1328	# Create a mapping table, each metric to the period where it should be stored
	1329	#
[53]	1330	metric_serial_table = self.determineSerials( hostname, metricname, metrics_to_store )
[33]	1331
[50]	1332	update_rets = [ ]
	1333
[47]	1334	for period, pmetric in metric_serial_table.items():
	1335
[146]	1336	create_ret = self.createCheck( hostname, metricname, period )
[47]	1337
	1338	update_ret = self.update( hostname, metricname, period, pmetric )
	1339
	1340	if update_ret == 0:
	1341
	1342	debug_msg( 9, 'stored metric %s for %s' %( hostname, metricname ) )
	1343	else:
	1344	debug_msg( 9, 'metric update failed' )
	1345
[146]	1346	update_rets.append( create_ret )
[50]	1347	update_rets.append( update_ret )
[47]	1348
[179]	1349	# Lets ignore errors here for now, we need to make sure last update time
	1350	# is correct!
	1351	#
	1352	#if not (1) in update_rets:
[50]	1353
[179]	1354	self.memLastUpdate( hostname, metricname, metrics_to_store )
[50]	1355
[17]	1356	def makeTimeSerial( self ):
[63]	1357	"""Generate a time serial. Seconds since epoch"""
[17]	1358
	1359	# Seconds since epoch
	1360	mytime = int( time.time() )
	1361
	1362	return mytime
	1363
[50]	1364	def makeRrdPath( self, host, metricname, timeserial ):
[63]	1365	"""Make a RRD location/path and filename"""
[17]	1366
[50]	1367	rrd_dir = '%s/%s/%s/%s' %( check_dir(ARCHIVE_PATH), self.cluster, host, timeserial )
	1368	rrd_file = '%s/%s.rrd' %( rrd_dir, metricname )
[17]	1369
	1370	return rrd_dir, rrd_file
	1371
[20]	1372	def getLastRrdTimeSerial( self, host ):
[63]	1373	"""Find the last timeserial (directory) for this host"""
[17]	1374
[19]	1375	newest_timeserial = 0
	1376
[47]	1377	for dir in self.timeserials[ host ]:
[32]	1378
[47]	1379	valid_dir = 1
[17]	1380
[47]	1381	for letter in dir:
	1382	if letter not in string.digits:
	1383	valid_dir = 0
[17]	1384
[47]	1385	if valid_dir:
	1386	timeserial = dir
	1387	if timeserial > newest_timeserial:
	1388	newest_timeserial = timeserial
[17]	1389
	1390	if newest_timeserial:
[18]	1391	return newest_timeserial
[17]	1392	else:
	1393	return 0
	1394
[47]	1395	def determinePeriod( self, host, check_serial ):
[63]	1396	"""Determine to which period (directory) this time(serial) belongs"""
[47]	1397
	1398	period_serial = 0
	1399
[56]	1400	if self.timeserials.has_key( host ):
[47]	1401
[56]	1402	for serial in self.timeserials[ host ]:
[47]	1403
[56]	1404	if check_serial >= serial and period_serial < serial:
[47]	1405
[56]	1406	period_serial = serial
	1407
[47]	1408	return period_serial
	1409
	1410	def determineSerials( self, host, metricname, metriclist ):
	1411	"""
	1412	Determine the correct serial and corresponding rrd to store
	1413	for a list of metrics
	1414	"""
	1415
	1416	metric_serial_table = { }
	1417
	1418	for metric in metriclist:
	1419
	1420	if metric['name'] == metricname:
	1421
	1422	period = self.determinePeriod( host, metric['time'] )
	1423
	1424	archive_secs = ARCHIVE_HOURS_PER_RRD * (60 * 60)
	1425
[49]	1426	if (int( metric['time'] ) - int( period ) ) > archive_secs:
[47]	1427
	1428	# This one should get it's own new period
	1429	period = metric['time']
[57]	1430
	1431	if not self.timeserials.has_key( host ):
	1432	self.timeserials[ host ] = [ ]
	1433
[50]	1434	self.timeserials[ host ].append( period )
[47]	1435
	1436	if not metric_serial_table.has_key( period ):
	1437
[49]	1438	metric_serial_table[ period ] = [ ]
[47]	1439
	1440	metric_serial_table[ period ].append( metric )
	1441
	1442	return metric_serial_table
	1443
[33]	1444	def createCheck( self, host, metricname, timeserial ):
[63]	1445	"""Check if an rrd allready exists for this metric, create if not"""
[9]	1446
[35]	1447	debug_msg( 9, 'rrdcreate: using timeserial %s for %s/%s' %( timeserial, host, metricname ) )
[47]	1448
[33]	1449	rrd_dir, rrd_file = self.makeRrdPath( host, metricname, timeserial )
[17]	1450
[9]	1451	if not os.path.exists( rrd_dir ):
[58]	1452
	1453	try:
	1454	os.makedirs( rrd_dir )
	1455
[169]	1456	except os.OSError, msg:
[58]	1457
	1458	if msg.find( 'File exists' ) != -1:
	1459
	1460	# Ignore exists errors
	1461	pass
	1462
	1463	else:
	1464
	1465	print msg
	1466	return
	1467
[14]	1468	debug_msg( 9, 'created dir %s' %( str(rrd_dir) ) )
[9]	1469
[14]	1470	if not os.path.exists( rrd_file ):
[9]	1471
[33]	1472	interval = self.config.getInterval( self.cluster )
[47]	1473	heartbeat = 8 * int( interval )
[9]	1474
[37]	1475	params = [ ]
[12]	1476
[37]	1477	params.append( '--step' )
	1478	params.append( str( interval ) )
[12]	1479
[37]	1480	params.append( '--start' )
[47]	1481	params.append( str( int( timeserial ) - 1 ) )
[12]	1482
[37]	1483	params.append( 'DS:sum:GAUGE:%d:U:U' %heartbeat )
	1484	params.append( 'RRA:AVERAGE:0.5:1:%s' %(ARCHIVE_HOURS_PER_RRD * 240) )
[13]	1485
[37]	1486	self.rrdm.create( str(rrd_file), params )
	1487
[14]	1488	debug_msg( 9, 'created rrd %s' %( str(rrd_file) ) )
	1489
[47]	1490	def update( self, host, metricname, timeserial, metriclist ):
[63]	1491	"""
	1492	Update rrd file for host with metricname
	1493	in directory timeserial with metriclist
	1494	"""
[9]	1495
[35]	1496	debug_msg( 9, 'rrdupdate: using timeserial %s for %s/%s' %( timeserial, host, metricname ) )
[9]	1497
[33]	1498	rrd_dir, rrd_file = self.makeRrdPath( host, metricname, timeserial )
[18]	1499
[47]	1500	update_list = self.makeUpdateList( host, metriclist )
[15]	1501
[41]	1502	if len( update_list ) > 0:
	1503	ret = self.rrdm.update( str(rrd_file), update_list )
[32]	1504
[41]	1505	if ret:
	1506	return 1
[27]	1507
[41]	1508	debug_msg( 9, 'updated rrd %s with %s' %( str(rrd_file), string.join( update_list ) ) )
[15]	1509
[36]	1510	return 0
	1511
[169]	1512	def daemon():
	1513	"""daemonized threading"""
[8]	1514
[169]	1515	# Fork the first child
	1516	#
	1517	pid = os.fork()
	1518
	1519	if pid > 0:
	1520
	1521	sys.exit(0) # end parent
	1522
	1523	# creates a session and sets the process group ID
	1524	#
	1525	os.setsid()
	1526
	1527	# Fork the second child
	1528	#
	1529	pid = os.fork()
	1530
	1531	if pid > 0:
	1532
	1533	sys.exit(0) # end parent
	1534
	1535	# Go to the root directory and set the umask
	1536	#
	1537	os.chdir('/')
	1538	os.umask(0)
	1539
	1540	sys.stdin.close()
	1541	sys.stdout.close()
	1542	sys.stderr.close()
	1543
[273]	1544	os.open('/dev/null', os.O_RDWR)
	1545	os.dup2(0, 1)
	1546	os.dup2(0, 2)
[169]	1547
	1548	run()
	1549
	1550	def run():
	1551	"""Threading start"""
	1552
[287]	1553	myXMLSource = XMLGatherer( ARCHIVE_XMLSOURCE.split( ':' )[0], ARCHIVE_XMLSOURCE.split( ':' )[1] )
[8]	1554
[287]	1555	myTorqueProcessor = TorqueXMLProcessor( myXMLSource )
	1556	myGangliaProcessor = GangliaXMLProcessor( myXMLSource )
	1557
[169]	1558	try:
[102]	1559	torque_xml_thread = threading.Thread( None, myTorqueProcessor.run, 'torque_proc_thread' )
	1560	ganglia_xml_thread = threading.Thread( None, myGangliaProcessor.run, 'ganglia_proc_thread' )
[22]	1561
[169]	1562	torque_xml_thread.start()
	1563	ganglia_xml_thread.start()
	1564
[176]	1565	except thread.error, msg:
[169]	1566	debug_msg( 0, 'FATAL ERROR: Unable to start main threads!: '+ str(msg) )
	1567	syslog.closelog()
	1568	sys.exit(1)
	1569
	1570	debug_msg( 0, 'main threading started.' )
[78]	1571
[169]	1572	def main():
	1573	"""Program startup"""
	1574
[214]	1575	if not processArgs( sys.argv[1:] ):
	1576	sys.exit( 1 )
	1577
[169]	1578	if( DAEMONIZE and USE_SYSLOG ):
	1579	syslog.openlog( 'jobarchived', syslog.LOG_NOWAIT, SYSLOG_FACILITY )
	1580
	1581	if DAEMONIZE:
	1582	daemon()
	1583	else:
	1584	run()
	1585
	1586	#
[81]	1587	# Global functions
[169]	1588	#
[81]	1589
[9]	1590	def check_dir( directory ):
[63]	1591	"""Check if directory is a proper directory. I.e.: Does _not_ end with a '/'"""
[9]	1592
	1593	if directory[-1] == '/':
	1594	directory = directory[:-1]
	1595
	1596	return directory
	1597
[12]	1598	def debug_msg( level, msg ):
[169]	1599	"""Only print msg if correct levels"""
[12]	1600
[169]	1601	if (not DAEMONIZE and DEBUG_LEVEL >= level):
	1602	sys.stderr.write( printTime() + ' - ' + msg + '\n' )
	1603
	1604	if (DAEMONIZE and USE_SYSLOG and SYSLOG_LEVEL >= level):
	1605	syslog.syslog( msg )
[12]	1606
[46]	1607	def printTime( ):
[63]	1608	"""Print current time in human readable format"""
[46]	1609
	1610	return time.strftime("%a %d %b %Y %H:%M:%S")
	1611
[63]	1612	# Ooohh, someone started me! Let's go..
[9]	1613	if __name__ == '__main__':
	1614	main()

Note: See TracBrowser for help on using the repository browser.

Download in other formats: