X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=findbad.py;h=9d2758c798fafa9e2d062cf121141110eb60040b;hb=6f2351e4b44590221425fa9b4bfa77c92db49b6a;hp=fa4c76a5c67b2346f87b4e9226004e0a9a9c8392;hpb=b93294f1bc40cdfe7f5e632577f8d19b24f97c37;p=monitor.git

diff --git a/findbad.py b/findbad.py
index fa4c76a..9d2758c 100755
--- a/findbad.py
+++ b/findbad.py
@@ -4,210 +4,284 @@ import os
 import sys
 import string
 import time
+from datetime import datetime,timedelta
+import threadpool
+import threading
+
+from monitor import util
+from monitor.util import command
+from monitor import config
+from monitor.database import FindbadNodeRecordSync, FindbadNodeRecord
+from monitor.sources import comon
+from monitor.wrapper import plc
 
-from config import config
-from optparse import OptionParser
-parser = OptionParser()
-parser.set_defaults(filename=None, increment=False, dbname="findbadnodes", cachenodes=False)
-parser.add_option("-f", "--nodelist", dest="filename", metavar="FILE", 
-					help="Provide the input file for the node list")
-parser.add_option("", "--cachenodes", action="store_true",
-					help="Cache node lookup from PLC")
-parser.add_option("", "--dbname", dest="dbname", metavar="FILE", 
-					help="Specify the name of the database to which the information is saved")
-parser.add_option("-i", "--increment", action="store_true", dest="increment", 
-					help="Increment round number to force refresh or retry")
-config = config(parser)
-config.parse_args()
+import syncplcdb
+from nodequery import verify,query_to_dict,node_select
+import traceback
 
+print "starting sqlfindbad.py"
 # QUERY all nodes.
 COMON_COTOPURL= "http://summer.cs.princeton.edu/status/tabulator.cgi?" + \
-					"table=table_nodeview&" + \
-				    "dumpcols='name,resptime,sshstatus,uptime,lastcotop'&" + \
-				    "formatcsv"
+				"table=table_nodeview&" + \
+				"dumpcols='name,resptime,sshstatus,uptime,lastcotop,cpuspeed,memsize,disksize'&" + \
+				"formatcsv"
 				    #"formatcsv&" + \
 					#"select='lastcotop!=0'"
 
-import threading
+api = plc.getAuthAPI()
 plc_lock = threading.Lock()
 round = 1
-externalState = {'round': round, 'nodes': {}}
+global_round = round
 count = 0
 
-
-import soltesz
-import plc
-import comon
-import threadpool
-import syncplcdb
-
 def collectPingAndSSH(nodename, cohash):
 	### RUN PING ######################
-	ping = soltesz.CMD()
-	(oval,eval) = ping.run_noexcept("ping -c 1 -q %s | grep rtt" % nodename)
-
-	values = {}
-
-	if oval == "":
-		# An error occurred
-		values['ping'] = "NOPING"
-	else:
-		values['ping'] = "PING"
-
-	#uptime = soltesz.SSH('root', nodename)
-	#(oval,eval) = uptime.run_noexcept("uptime | awk '{print $3,$4}' | tr , ' '")
-
-	### RUN SSH ######################
-	b_getbootcd_id = True
-	ssh = soltesz.SSH('root', nodename)
-	oval = ""
-	eval = ""
-	(oval, eval) = ssh.run_noexcept('echo `uname -a ; ls /tmp/bm.log`')
-	val = oval
-	if "2.6.17" in oval or "2.6.2" in oval:
-		values['ssh'] = 'SSH'
-		values['category'] = 'ALPHA'
-		if "bm.log" in oval:
-			values['state'] = 'DEBUG'
-		else:
-			values['state'] = 'BOOT'
-	elif "2.6.12" in oval or "2.6.10" in oval:
-		values['ssh'] = 'SSH'
-		values['category'] = 'PROD'
-		if "bm.log" in oval:
-			values['state'] = 'DEBUG'
+	ping = command.CMD()
+	(oval,errval) = ping.run_noexcept("ping -c 1 -q %s | grep rtt" % nodename)
+
+	try:
+		values = {}
+
+		if oval == "":
+			# An error occurred
+			values['ping'] = "NOPING"
 		else:
-			values['state'] = 'BOOT'
-	elif "2.4" in oval:
-		b_getbootcd_id = False
-		values['ssh'] = 'SSH'
-		values['category'] = 'OLDBOOTCD'
-		values['state'] = 'DEBUG'
-	elif oval != "":
-		values['ssh'] = 'SSH'
-		values['category'] = 'UNKNOWN'
-		if "bm.log" in oval:
+			values['ping'] = "PING"
+
+		try:
+			for port in [22, 806]: 
+				ssh = command.SSH('root', nodename, port)
+
+				(oval, errval) = ssh.run_noexcept2(""" <<\EOF
+					echo "{"
+					echo '  "kernel":"'`uname -a`'",'
+					echo '  "bmlog":"'`ls /tmp/bm.log`'",'
+					echo '  "bootcd":"'`cat /mnt/cdrom/bootme/ID`'",'
+					echo '  "nm":"'`ps ax | grep nm.py | grep -v grep`'",'
+					echo '  "readonlyfs":"'`touch /var/log/monitor 2>&1`'",'
+					echo '  "dns":"'`host boot.planet-lab.org 2>&1`'",'
+					echo '  "princeton_comon":"'`ls -d /vservers/princeton_comon`'",'
+
+					ID=`grep princeton_comon /etc/passwd | awk -F : '{if ( $3 > 500 ) { print $3}}'` 
+					echo '  "princeton_comon_running":"'`ls -d /proc/virtual/$ID`'",'
+					echo '  "princeton_comon_procs":"'`vps ax | grep $ID | grep -v grep | wc -l`'",'
+					echo "}"
+EOF				""")
+				
+				values['ssherror'] = errval
+				if len(oval) > 0:
+					#print "OVAL: %s" % oval
+					values.update(eval(oval))
+					values['sshport'] = port
+					break
+				else:
+					values.update({'kernel': "", 'bmlog' : "", 'bootcd' : '', 
+									'nm' : '', 
+									'readonlyfs' : '',
+									'dns' : '',
+									'princeton_comon' : "", 
+									'princeton_comon_running' : "", 
+									'princeton_comon_procs' : "", 'sshport' : None})
+		except:
+			print traceback.print_exc()
+			sys.exit(1)
+
+		### RUN SSH ######################
+		b_getbootcd_id = True
+		#ssh = command.SSH('root', nodename)
+		#oval = ""
+		#errval = ""
+		#(oval, errval) = ssh.run_noexcept('echo `uname -a ; ls /tmp/bm.log`')
+
+		oval = values['kernel']
+		if "2.6.17" in oval or "2.6.2" in oval:
+			values['ssh'] = 'SSH'
+			values['category'] = 'PROD'
+			if "bm.log" in values['bmlog']:
+				values['state'] = 'DEBUG'
+			else:
+				values['state'] = 'BOOT'
+		elif "2.6.12" in oval or "2.6.10" in oval:
+			values['ssh'] = 'SSH'
+			values['category'] = 'OLDPROD'
+			if "bm.log" in values['bmlog']:
+				values['state'] = 'DEBUG'
+			else:
+				values['state'] = 'BOOT'
+		
+		# NOTE: on 2.6.8 kernels, with 4.2 bootstrapfs, the chroot command fails.  I have no idea why.
+		elif "2.4" in oval or "2.6.8" in oval:
+			b_getbootcd_id = False
+			values['ssh'] = 'SSH'
+			values['category'] = 'OLDBOOTCD'
 			values['state'] = 'DEBUG'
+		elif oval != "":
+			values['ssh'] = 'SSH'
+			values['category'] = 'UNKNOWN'
+			if "bm.log" in values['bmlog']:
+				values['state'] = 'DEBUG'
+			else:
+				values['state'] = 'BOOT'
 		else:
-			values['state'] = 'BOOT'
-	else:
-		# An error occurred.
-		b_getbootcd_id = False
-		values['ssh'] = 'NOSSH'
-		values['category'] = 'ERROR'
-		values['state'] = 'DOWN'
-		val = eval.strip()
-
-	values['kernel'] = val
-
-	if b_getbootcd_id:
-		# try to get BootCD for all nodes that are not 2.4 nor inaccessible
-		(oval, eval) = ssh.run_noexcept('cat /mnt/cdrom/bootme/ID')
-		val = oval
-		if "BootCD" in val:
-			values['bootcd'] = val
-			if "v2" in val and \
-				( nodename is not "planetlab1.cs.unc.edu" and \
-				  nodename is not "planetlab2.cs.unc.edu" ):
-				values['category'] = 'OLDBOOTCD'
+			# An error occurred.
+			b_getbootcd_id = False
+			values['ssh'] = 'NOSSH'
+			values['category'] = 'ERROR'
+			values['state'] = 'DOWN'
+			val = errval.strip()
+			values['ssherror'] = val
+			values['kernel'] = ""
+
+		#values['kernel'] = val
+
+		if b_getbootcd_id:
+			# try to get BootCD for all nodes that are not 2.4 nor inaccessible
+			#(oval, errval) = ssh.run_noexcept('cat /mnt/cdrom/bootme/ID')
+			oval = values['bootcd']
+			if "BootCD" in oval:
+				values['bootcd'] = oval
+				if "v2" in oval and \
+					( nodename is not "planetlab1.cs.unc.edu" and \
+					  nodename is not "planetlab2.cs.unc.edu" ):
+					values['category'] = 'OLDBOOTCD'
+			else:
+				values['bootcd'] = ""
 		else:
 			values['bootcd'] = ""
-	else:
-		values['bootcd'] = ""
 
-	# TODO: get bm.log for debug nodes.
-	# 'zcat /tmp/bm.log'
+		# TODO: get bm.log for debug nodes.
+		# 'zcat /tmp/bm.log'
 		
-	if nodename in cohash: 
-		values['comonstats'] = cohash[nodename]
-	else:
-		values['comonstats'] = {'resptime':  '-1', 
-								'uptime':    '-1',
-								'sshstatus': '-1', 
-								'lastcotop': '-1'}
-	# include output value
-	### GET PLC NODE ######################
-	b_except = False
-	plc_lock.acquire()
-
-	try:
-		d_node = plc.getNodes({'hostname': nodename}, ['pcu_ids', 'site_id', 'last_contact', 'boot_state', 'nodegroup_ids'])
-	except:
-		b_except = True
-		import traceback
-		b_except = True
-		import traceback
-		traceback.print_exc()
-
-	plc_lock.release()
-	if b_except: return (None, None)
-
-	site_id = -1
-	if d_node and len(d_node) > 0:
-		pcu = d_node[0]['pcu_ids']
-		if len(pcu) > 0:
-			values['pcu'] = "PCU"
+		#(oval, errval) = ssh.run_noexcept('ps ax | grep nm.py | grep -v grep')
+		oval = values['nm']
+		if "nm.py" in oval:
+			values['nm'] = "Y"
 		else:
-			values['pcu'] = "NOPCU"
-		site_id = d_node[0]['site_id']
-		last_contact = d_node[0]['last_contact']
-		nodegroups = d_node[0]['nodegroup_ids']
-		values['plcnode'] = {'status' : 'SUCCESS', 
-							'pcu_ids': pcu, 
-							'boot_state' : d_node[0]['boot_state'],
-							'site_id': site_id,
-							'nodegroups' : nodegroups,
-							'last_contact': last_contact}
-	else:
-		values['pcu']     = "UNKNOWN"
-		values['plcnode'] = {'status' : "GN_FAILED"}
-		
+			values['nm'] = "N"
 
-	### GET PLC SITE ######################
-	b_except = False
-	plc_lock.acquire()
+		continue_slice_check = True
+		#(oval, errval) = ssh.run_noexcept('ls -d /vservers/princeton_comon')
+		oval = values['princeton_comon']
+		if "princeton_comon" in oval:
+			values['princeton_comon'] = True
+		else:
+			values['princeton_comon'] = False
+			continue_slice_check = False
+
+		if continue_slice_check:
+			#(oval, errval) = ssh.run_noexcept('ID=`grep princeton_comon /etc/passwd | awk -F : "{if ( \\\$3 > 500 ) { print \\\$3}}"`; ls -d /proc/virtual/$ID')
+			oval = values['princeton_comon_running']
+			if len(oval) > len('/proc/virtual/'):
+				values['princeton_comon_running'] = True
+			else:
+				values['princeton_comon_running'] = False
+				continue_slice_check = False
+		else:
+			values['princeton_comon_running'] = False
+			
+		if continue_slice_check:
+			#(oval, errval) = ssh.run_noexcept('ID=`grep princeton_comon /etc/passwd | awk -F : "{if ( \\\$3 > 500 ) { print \\\$3}}"`; vps ax | grep $ID | grep -v grep | wc -l')
+			oval = values['princeton_comon_procs']
+			values['princeton_comon_procs'] = int(oval)
+		else:
+			values['princeton_comon_procs'] = None
 
-	try:
-		d_site = plc.getSites({'site_id': site_id}, 
-							['max_slices', 'slice_ids', 'node_ids', 'login_base'])
+			
+		if nodename in cohash: 
+			values['comonstats'] = cohash[nodename]
+		else:
+			values['comonstats'] = {'resptime':  '-1', 
+									'uptime':    '-1',
+									'sshstatus': '-1', 
+									'lastcotop': '-1',
+									'cpuspeed' : "null",
+									'disksize' : 'null',
+									'memsize'  : 'null'}
+		# include output value
+		### GET PLC NODE ######################
+		plc_lock.acquire()
+		d_node = None
+		try:
+			d_node = plc.getNodes({'hostname': nodename}, ['pcu_ids', 'site_id', 'date_created', 
+									'last_updated', 'last_contact', 'boot_state', 'nodegroup_ids'])[0]
+		except:
+			traceback.print_exc()
+		plc_lock.release()
+		values['plcnode'] = d_node
+
+		### GET PLC PCU ######################
+		site_id = -1
+		d_pcu = None
+		if d_node:
+			pcu = d_node['pcu_ids']
+			if len(pcu) > 0:
+				d_pcu = pcu[0]
+
+			site_id = d_node['site_id']
+
+		values['pcu'] = d_pcu
+
+		### GET PLC SITE ######################
+		plc_lock.acquire()
+		d_site = None
+		values['loginbase'] = ""
+		try:
+			d_site = plc.getSites({'site_id': site_id}, 
+								['max_slices', 'slice_ids', 'node_ids', 'login_base'])[0]
+			values['loginbase'] = d_site['login_base']
+		except:
+			traceback.print_exc()
+		plc_lock.release()
+
+		values['plcsite'] = d_site 
+		values['date_checked'] = time.time()
 	except:
-		b_except = True
-		import traceback
-		traceback.print_exc()
-
-	plc_lock.release()
-	if b_except: return (None, None)
-
-	if d_site and len(d_site) > 0:
-		max_slices = d_site[0]['max_slices']
-		num_slices = len(d_site[0]['slice_ids'])
-		num_nodes = len(d_site[0]['node_ids'])
-		loginbase = d_site[0]['login_base']
-		values['plcsite'] = {'num_nodes' : num_nodes, 
-							'max_slices' : max_slices, 
-							'num_slices' : num_slices,
-							'login_base' : loginbase,
-							'status'     : 'SUCCESS'}
-	else:
-		values['plcsite'] = {'status' : "GS_FAILED"}
-
-	values['checked'] = time.time()
+		print traceback.print_exc()
 
 	return (nodename, values)
 
 def recordPingAndSSH(request, result):
-	global externalState
+	global global_round
 	global count
 	(nodename, values) = result
 
-	if values is not None:
-		global_round = externalState['round']
-		externalState['nodes'][nodename]['values'] = values
-		externalState['nodes'][nodename]['round'] = global_round
+	try:
+		if values is not None:
+			fbsync = FindbadNodeRecordSync.findby_or_create(hostname="global", 
+															if_new_set={'round' : global_round})
+			global_round = fbsync.round
+			fbnodesync = FindbadNodeRecordSync.findby_or_create(hostname=nodename,
+															if_new_set={'round' : global_round})
+
+			fbrec = FindbadNodeRecord(
+						date_checked=datetime.fromtimestamp(values['date_checked']),
+						hostname=nodename,
+						loginbase=values['loginbase'],
+						kernel_version=values['kernel'],
+						bootcd_version=values['bootcd'],
+						nm_status=values['nm'],
+						fs_status=values['readonlyfs'],
+						dns_status=values['dns'],
+						princeton_comon_dir=values['princeton_comon'],
+						princeton_comon_running=values['princeton_comon_running'],
+						princeton_comon_procs=values['princeton_comon_procs'],
+						plc_node_stats = values['plcnode'],
+						plc_site_stats = values['plcsite'],
+						plc_pcuid = values['pcu'],
+						comon_stats = values['comonstats'],
+						ping_status = (values['ping'] == "PING"),
+						ssh_portused = values['sshport'],
+						ssh_status = (values['ssh'] == "SSH"),
+						ssh_error = values['ssherror'],
+						observed_status = values['state'],
+					)
+			fbnodesync.round = global_round
 
-		count += 1
-		print "%d %s %s" % (count, nodename, externalState['nodes'][nodename]['values'])
-		soltesz.dbDump(config.dbname, externalState)
+			count += 1
+			print "%d %s %s" % (count, nodename, values)
+	except:
+		print "ERROR:"
+		print traceback.print_exc()
 
 # this will be called when an exception occurs within a thread
 def handle_exception(request, result):
@@ -217,18 +291,16 @@ def handle_exception(request, result):
 
 
 def checkAndRecordState(l_nodes, cohash):
-	global externalState
+	global global_round
 	global count
-	global_round = externalState['round']
 
 	tp = threadpool.ThreadPool(20)
 
 	# CREATE all the work requests
 	for nodename in l_nodes:
-		if nodename not in externalState['nodes']:
-			externalState['nodes'][nodename] = {'round': 0, 'values': []}
+		fbnodesync = FindbadNodeRecordSync.findby_or_create(hostname=nodename, if_new_set={'round':0})
 
-		node_round   = externalState['nodes'][nodename]['round']
+		node_round   = fbnodesync.round
 		if node_round < global_round:
 			# recreate node stats when refreshed
 			#print "%s" % nodename
@@ -238,14 +310,19 @@ def checkAndRecordState(l_nodes, cohash):
 		else:
 			# We just skip it, since it's "up to date"
 			count += 1
-			print "%d %s %s" % (count, nodename, externalState['nodes'][nodename]['values'])
-			pass
+			#print "%d %s %s" % (count, nodename, externalState['nodes'][nodename]['values'])
+			print "%d %s %s" % (count, nodename, node_round)
 
 	# WAIT while all the work requests are processed.
+	begin = time.time()
 	while 1:
 		try:
 			time.sleep(1)
 			tp.poll()
+			# if more than two hours
+			if time.time() - begin > (60*60*1.5):
+				print "findbad.py has run out of time!!!!!!"
+				os._exit(1)
 		except KeyboardInterrupt:
 			print "Interrupted!"
 			break
@@ -253,16 +330,20 @@ def checkAndRecordState(l_nodes, cohash):
 			print "All results collected."
 			break
 
-
+	print FindbadNodeRecordSync.query.count()
+	print FindbadNodeRecord.query.count()
 
 def main():
-	global externalState
+	global global_round
 
-	externalState = soltesz.if_cached_else(1, config.dbname, lambda : externalState) 
+	fbsync = FindbadNodeRecordSync.findby_or_create(hostname="global", 
+													if_new_set={'round' : global_round})
+	global_round = fbsync.round
 
 	if config.increment:
 		# update global round number to force refreshes across all nodes
-		externalState['round'] += 1
+		global_round += 1
+		fbsync.round = global_round
 
 	cotop = comon.Comon()
 	# lastcotop measures whether cotop is actually running.  this is a better
@@ -270,14 +351,31 @@ def main():
 	cotop_url = COMON_COTOPURL
 
 	# history information for all nodes
+	#cohash = {}
 	cohash = cotop.coget(cotop_url)
 	l_nodes = syncplcdb.create_plcdb()
-	if config.filename:
-		f_nodes = config.getListFromFile(config.filename)
+	if config.nodelist:
+		f_nodes = util.file.getListFromFile(config.nodelist)
 		l_nodes = filter(lambda x: x['hostname'] in f_nodes, l_nodes)
-
+	elif config.node:
+		f_nodes = [config.node]
+		l_nodes = filter(lambda x: x['hostname'] in f_nodes, l_nodes)
+	elif config.nodegroup:
+		ng = api.GetNodeGroups({'name' : config.nodegroup})
+		l_nodes = api.GetNodes(ng[0]['node_ids'])
+	elif config.site:
+		site = api.GetSites(config.site)
+		l_nodes = api.GetNodes(site[0]['node_ids'], ['hostname'])
+		
 	l_nodes = [node['hostname'] for node in l_nodes]
 
+	# perform this query after the above options, so that the filter above
+	# does not break.
+	if config.nodeselect:
+		plcnodes = api.GetNodes({'peer_id' : None}, ['hostname'])
+		plcnodes = [ node['hostname'] for node in plcnodes ]
+		l_nodes = node_select(config.nodeselect, plcnodes, None)
+
 	print "fetching %s hosts" % len(l_nodes)
 
 	checkAndRecordState(l_nodes, cohash)
@@ -286,12 +384,29 @@ def main():
 
 
 if __name__ == '__main__':
+	from monitor import parser as parsermodule
+
+	parser = parsermodule.getParser(['nodesets'])
+
+	parser.set_defaults( increment=False, dbname="findbad", cachenodes=False)
+	parser.add_option("", "--cachenodes", action="store_true",
+						help="Cache node lookup from PLC")
+	parser.add_option("", "--dbname", dest="dbname", metavar="FILE", 
+						help="Specify the name of the database to which the information is saved")
+	parser.add_option("-i", "--increment", action="store_true", dest="increment", 
+						help="Increment round number to force refresh or retry")
+
+	parser = parsermodule.getParser(['defaults'], parser)
+	
+	cfg = parsermodule.parse_args(parser)
+
 	try:
 		main()
 	except Exception, err:
-		import traceback
 		print traceback.print_exc()
 		print "Exception: %s" % err
 		print "Saving data... exitting."
-		soltesz.dbDump(config.dbname, externalState)
 		sys.exit(0)
+	print "sleeping"
+	#print "final commit"
+	#time.sleep(10)