def checkProcesses( liveInstances ):
'''check that the expected process is running on each instance'''
logger.info( 'checking %d instance(s)', len(liveInstances) )
#maybe should weed out some instances
goodInstances = liveInstances
cmd = "ps -ef | grep -v grep | grep 'agent.jar' > /dev/null"
#cmd = "free --mega 1>&2 ; ps -ef | grep -v grep | grep 'LoadGeneratorAgent start' > /dev/null"
# check for a running agent process on each instance
stepStatuses = tellInstances.tellInstances( goodInstances, cmd,
timeLimit=30*60,
knownHostsOnly=True
)
#logger.info( 'cmd statuses: %s', stepStatuses )
errorsByIid = {}
for statusRec in stepStatuses:
status = statusRec['status']
if status:
logger.info( 'statusRec: %s', statusRec )
# only certain non-null outcomes are true errors
if isinstance( status, Exception ):
errorsByIid[ statusRec['instanceId'] ] = statusRec
elif status not in [0, 1]:
errorsByIid[ statusRec['instanceId'] ] = statusRec
logger.info( 'errorsByIid: %s', errorsByIid )
return errorsByIid
def retrieveLogs( liveInstances, neoloadVersion ):
'''
instancesFilePath = os.path.join( dataDirPath, 'startedAgents.json' )
startedInstances = []
# get details of launched instances from the json file
#TODO should get list of instances with good install, rather than all started instances
with open( instancesFilePath, 'r') as jsonInFile:
try:
startedInstances = json.load(jsonInFile) # an array
except Exception as exc:
logger.warning( 'could not load json (%s) %s', type(exc), exc )
'''
#maybe should weed out some instances
goodInstances = liveInstances
logger.info( 'retrieving logs for %d instance(s)', len(goodInstances) )
truncVersion = truncateVersion( neoloadVersion )
agentLogFilePath = '/root/.neotys/neoload/v%s/logs/*.log' % truncVersion
# download the agent.log file from each instance
stepStatuses = tellInstances.tellInstances( goodInstances,
download=agentLogFilePath, downloadDestDir=dataDirPath +'/agentLogs',
timeLimit=30*60,
knownHostsOnly=True
)
def retrieveLogs( goodInstances ):
'''download proxy log files from each of the given instances '''
logger.info( 'retrieving logs for %d instance(s)', len(goodInstances) )
proxyLogFilePath = '/var/log/squid/*.log'
stepStatuses = tellInstances.tellInstances( goodInstances,
download=proxyLogFilePath, downloadDestDir=dataDirPath +'/proxyLogs',
timeLimit=30*60,
knownHostsOnly=True
)
def retrieveLogs( goodInstances ):
logger.info( 'retrieving logs for %d instance(s)', len(goodInstances) )
#agentLogFilePath = 'lzAgent/Logs/*'
agentLogFilePath = 'lzAgent/agent.*s'
# download the agent.log file from each instance
stepStatuses = tellInstances.tellInstances( goodInstances,
download=agentLogFilePath, downloadDestDir=dataDirPath +'/agentLogs',
timeLimit=30*60,
knownHostsOnly=True
)
def retrieveLogs( liveInstances, farmDirPath ):
'''retrieve logs from nodes, storing them in dataDirPath/nodeLogs subdirectories'''
goodInstances = liveInstances
logger.info( 'retrieving logs for %d instance(s)', len(goodInstances) )
nodeLogFilePath = farmDirPath + '/geth.log'
# download the log file from each instance
stepStatuses = tellInstances.tellInstances( goodInstances,
download=nodeLogFilePath, downloadDestDir=dataDirPath +'/nodeLogs',
timeLimit=15*60,
knownHostsOnly=True
)
#logger.info( 'download statuses: %s', stepStatuses )
errorsByIid = {status['instanceId']: status['status'] for status in stepStatuses if status['status'] }
return errorsByIid
def checkProcesses( liveInstances ):
logger.info( 'checking %d instance(s)', len(liveInstances) )
#maybe should weed out some instances
goodInstances = liveInstances
cmd = "ps -ef | grep -v grep | grep 'geth' > /dev/null"
# check for a running geth process on each instance
stepStatuses = tellInstances.tellInstances( goodInstances, cmd,
timeLimit=15*60,
resultsLogFilePath = dataDirPath + '/checkProcesses.jlog',
knownHostsOnly=True
)
#logger.info( 'proc statuses: %s', stepStatuses )
errorsByIid = {status['instanceId']: status['status'] for status in stepStatuses if status['status'] }
#logger.info( 'errorsByIid: %s', errorsByIid )
return errorsByIid
def checkProcesses( goodInstances ):
'''check that the expected process is running on each instance'''
logger.info( 'checking %d instance(s)', len(goodInstances) )
cmd = "ps -ef | grep -v grep | grep 'squid' > /dev/null"
stepStatuses = tellInstances.tellInstances( goodInstances, cmd,
timeLimit=30*60,
knownHostsOnly=True
)
#logger.info( 'cmd statuses: %s', stepStatuses )
errorsByIid = {}
for status in stepStatuses:
if status['status']:
# might like to be more selective here
#logger.info( 'status: %s', status )
errorsByIid[ status['instanceId'] ] = status
logger.info( 'errorsByIid: %s', errorsByIid )
return errorsByIid
def checkInstanceClocks( liveInstances, dataDirPath ):
jlogFilePath = dataDirPath + '/checkInstanceClocks.jlog'
allIids = [inst['instanceId'] for inst in liveInstances ]
unfoundIids = set( allIids )
cmd = "date --iso-8601=seconds"
# check for a running geth process on each instance
stepStatuses = tellInstances.tellInstances( liveInstances, cmd,
timeLimit=2*60,
resultsLogFilePath = jlogFilePath,
knownHostsOnly=True, sshAgent=not True
)
#logger.info( 'proc statuses: %s', stepStatuses )
errorsByIid = {status['instanceId']: status['status'] for status in stepStatuses if status['status'] }
logger.info( 'preliminary errorsByIid: %s', errorsByIid )
for iid, status in errorsByIid.items():
logger.warning( 'instance %s gave error "%s"', iid, status )
with open( jlogFilePath, 'rb' ) as inFile:
for line in inFile:
decoded = json.loads( line )
iid = decoded['instanceId']
if decoded.get( 'stdout' ):
#logger.info( decoded )
masterDateTime = dateutil.parser.parse( decoded['dateTime'] )
try:
nodeDateTime = dateutil.parser.parse( decoded['stdout'] )
except Exception as exc:
logger.warning( 'exception parsing %s', decoded['stdout'] )
errorsByIid[ iid ] = {'exception': exc }
else:
unfoundIids.discard( iid )
delta = masterDateTime - nodeDateTime
discrep =delta.total_seconds()
logger.info( 'discrep: %.1f seconds on inst %s',
discrep, iid )
if discrep > 4 or discrep < -1:
logger.warning( 'bad time discrep: %.1f', discrep )
errorsByIid[ iid ] = {'discrep': discrep }
if unfoundIids:
logger.warning( 'unfoundIids: %s', unfoundIids )
for iid in list( unfoundIids ):
if iid not in errorsByIid:
errorsByIid[ iid ] = {'found': False }
logger.info( '%d errorsByIid: %s', len(errorsByIid), errorsByIid )
return errorsByIid
def checkProcesses( liveInstances ):
'''check that the expected process is running on each instance'''
logger.info( 'checking %d instance(s)', len(liveInstances) )
#maybe should weed out some instances
goodInstances = liveInstances
cmd = "ps -ef | grep -v grep | grep 'LoadGeneratorAgent start' > /dev/null"
#cmd = "free --mega 1>&2 ; ps -ef | grep -v grep | grep 'LoadGeneratorAgent start' > /dev/null"
# check for a running agent process on each instance
stepStatuses = tellInstances.tellInstances( goodInstances, cmd,
timeLimit=30*60,
knownHostsOnly=True
)
#logger.info( 'cmd statuses: %s', stepStatuses )
errorsByIid = {}
for status in stepStatuses:
if status['status']:
# might like to be more selective here
#logger.info( 'status: %s', status )
errorsByIid[ status['instanceId'] ] = status
logger.debug( 'errorsByIid: %s', errorsByIid )
return errorsByIid
with open(launchedJsonFilePath, 'r') as jsonInFile:
try:
launchedInstances = json.load(jsonInFile) # an array
except Exception as exc:
logger.warning('could not load json (%s) %s', type(exc), exc)
startedInstances = [
inst for inst in launchedInstances if inst['state'] == 'started'
]
logger.info('%d instances were launched', len(startedInstances))
starterCmd = 'geth --config netconfig/%s.config.toml --password pw.txt --unlock $(cat accountAddr.txt) >> ether/%s/stdout.txt 2>>ether/%s/geth.log </dev/null &' % \
(configName, configName, configName)
# start the client on each instance
stepStatuses = tellInstances.tellInstances(
startedInstances,
command=starterCmd,
resultsLogFilePath=outDataDir + '/startClients.jlog',
timeLimit=30 * 60,
knownHostsOnly=True)
logger.debug('starter statuses: %s', stepStatuses)
# make a list of instances where the client was started
goodIids = []
for status in stepStatuses:
if isinstance(status['status'], int) and status['status'] == 0:
goodIids.append(status['instanceId'])
else:
logger.warning('could not start agent on %s',
status['instanceId'][0:8])
if launchedInstances:
print(
'when you want to terminate these instances, use %s terminateGethNodes.py "%s"'
% (sys.executable, outDataDir))
startedInstances = [inst for inst in launchedInstances if inst['state'] == 'started' ]
logger.info( '%d instances were launched', len(startedInstances) )
installedInstances = []
recruiterLog = readJLog( outDataDir +'/recruitInstances.jlog' )
for logEntry in recruiterLog:
if 'returncode' in logEntry and logEntry['returncode'] == 0:
installedInstances.append( instancesByIid[ logEntry['instanceId'] ] )
logger.info( '%d instances were installed', len(installedInstances) )
installedIids = [inst['instanceId'] for inst in installedInstances ]
# get the ip addr of each instance
cmd = 'curl -s -S https://api.ipify.org > ipAddr.txt'
stepStatuses = tellInstances.tellInstances( installedInstances, command=cmd,
resultsLogFilePath=outDataDir +'/getIpAddr.jlog',
download='ipAddr.txt', downloadDestDir=outDataDir +'/agentLogs',
timeLimit=6*60,
knownHostsOnly=True
)
logger.debug( 'download statuses: %s', stepStatuses )
# merge the retrieved ip addrs into instances records in a new array
goodInstances = []
for iid in installedIids:
ipFilePath = os.path.join( outDataDir, 'agentLogs', iid, 'ipAddr.txt' )
if os.path.isfile( ipFilePath ):
with open( ipFilePath, 'r' ) as ipFile:
ipAddr = ipFile.read().strip()
if ipAddr:
inst = instancesByIid.get( iid )
if inst:
inst['ipAddr'] = ipAddr
for index, inst in enumerate( startedInstances ):
iid = inst['instanceId']
portMap[iid] = index + portRangeStart
logger.info( 'configuring agents')
returnCodes = configureAgents( startedInstances, ports, timeLimit=600 )
for index, code in enumerate( returnCodes ):
if code==0:
configuredInstances.append( startedInstances[index] )
else:
iid = startedInstances[index].get('instanceId')
logger.info( 'inst %s was not configured properly', iid[0:8] )
# start the agent on each instance
stepStatuses = tellInstances.tellInstances( configuredInstances, command=starterCmd,
resultsLogFilePath=outDataDir +'/startAgents.jlog',
timeLimit=30*60,
knownHostsOnly=True
)
logger.debug( 'starter statuses: %s', stepStatuses )
# make a list of instances where the agent was started
goodIids = []
for status in stepStatuses:
if isinstance( status['status'], int) and status['status'] == 0:
goodIids.append( status['instanceId'])
else:
logger.warning( 'could not start agent on %s', status['instanceId'][0:8] )
#COULD check bound ports again here
#COULD download logs from all installed instances rather than just good-started instances
goodInstances = [inst for inst in startedInstances if inst['instanceId'] in goodIids ]
if goodInstances:
time.sleep( 60 )
