Blame view
GstarCluster.py
11.7 KB
2382daaac
first commit
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 |
import paramiko,time
import Utils
from pysqlite2 import dbapi2 as sqlite
import pexpect, os, shutil, fileinput, sys, logging
class GstarCluster(object):
'''
This class holds all nodes of the db in the virtual cluster. It can start/stop individual
daemons as needed, thus adding/removing nodes at will. It also sets up the configuration
files as needed.
'''
def __init__(self, properties_file, initial_cluster_id = "default"):
'''
Constructor
'''
## Necessary variables
self.cluster = {}
self.host_template = ""
self.cluster_id = initial_cluster_id
self.utils = Utils.Utils(properties_file)
# Make sure the sqlite file exists. if not, create it and add the table we need
con = sqlite.connect(self.utils.db_file)
cur = con.cursor()
try:
clusters = cur.execute('select * from clusters',
).fetchall()
if len(clusters) > 0 :
print """Already discovered cluster id from previous database file. Will select the defined one to work with (if it exists)."""
# print "Found records:
", clusters
clustersfromcid = cur.execute('select * from clusters where cluster_id=\"'+self.cluster_id+"\"",
).fetchall()
if len(clustersfromcid) > 0 :
self.cluster = self.utils.get_cluster_from_db(self.cluster_id)
# print self.cluster
for clusterkey in self.cluster.keys():
if not (clusterkey.find("master") == -1):
self.host_template = clusterkey.replace("master","")
# Add self to db (eliminates existing records of same id)
self.utils.add_to_cluster_db(self.cluster, self.cluster_id)
else:
print "No known cluster with this id - run configure before you proceed"
except sqlite.DatabaseError:
cur.execute('create table clusters(cluster_id text, hostname text, ip_address text)')
con.commit()
cur.close()
con.close()
## Install logger
LOG_FILENAME = self.utils.install_dir+'/'+self.utils.logs_folder_name+'/Coordinator.log'
self.my_logger = logging.getLogger('GstarCluster')
self.my_logger.setLevel(self.utils.logging_level)
handler = logging.handlers.RotatingFileHandler(
LOG_FILENAME, maxBytes=2*1024*1024, backupCount=5)
formatter = logging.Formatter("%(asctime)s - %(levelname)s - %(name)s - %(message)s")
handler.setFormatter(formatter)
self.my_logger.addHandler(handler)
# def describe_nodes(self)
def configure_cluster(self, nodes=None, host_template=""):
## Check installation and print errors for nodes that do not exist/
## can not connect/have incorrect installed versions and/or paths
#nodes = self.block_until_ping(nodes)
hosts = open('/tmp/hosts', 'w')
i = 0
hosts.write("127.0.0.1\tlocalhost
")
# Connect to client2 in order to update the /etc/hosts file so that new nodes will be sent requests
#ssh_client2= paramiko.SSHClient()
#ssh_client2.set_missing_host_key_policy(paramiko.AutoAddPolicy())
#self.my_logger.debug("Connecting to client2")
#ssh_client2.connect('2001:648:2000:3:16:3eff:fe01:269a', username='root', password='secretpw')
for node in nodes:
ssh = paramiko.SSHClient()
ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
self.my_logger.debug("Starting config for node: " + node.ip_address)
ssh.connect(node.ip_address, username='root', password='secretpw')
## Check for installation dirs, otherwise exit with error message
stderr_all = []
stdin, stdout, stderr = ssh.exec_command('echo "root - nofile 200000" >> /etc/security/limits.conf')
stderr_all.append(stderr.readlines())
#stdin, stdout, stderr = ssh.exec_command('swapoff -a -v')
for stderr in stderr_all:
if len(stderr) > 0 :
self.my_logger.debug("ERROR - some installation files are missing")
return
stdin, stdout, stderr = ssh.exec_command('mkdir /root/.ssh')
if i==0:
# Add the master to the /etc/hosts file
hosts.write(node.ip_address + "\t" + host_template+"master
")
# Set hostname on the machine
stdin, stdout, stderr = ssh.exec_command('echo \"'+host_template+"master\" > /etc/hostname")
stdin, stdout, stderr = ssh.exec_command('hostname \"'+host_template+"master\"")
ssh.exec_command('echo "JAVA_OPTS=\"-Xms'+str(int(int(self.utils.m_ram)*0.9))+'m -Xmx'+str(int(int(self.utils.m_ram)*0.9))+'m\"" >> /root/.bashrc')
ssh.exec_command('source /root/.bashrc')
# Add node to cluster
self.cluster[host_template+"master"] = node
# Mount datasets dir
stdin, stdout, stderr = ssh.exec_command('mkdir /media/datasets')
stdin, stdout, stderr = ssh.exec_command("echo '192.168.1.4:/media/CloudImages/datasets /media/datasets nfs rw,sync,hard,intr 0 0' >> /etc/fstab")
stdin, stdout, stderr = ssh.exec_command('mount /media/datasets')
# sending experiment execution files
stdin, stdout, stderr = ssh.exec_command('mkdir /root/gstar')
transport = paramiko.Transport((node.ip_address, 22))
transport.connect(username = 'root', password = 'secretpw')
transport.open_channel("session", node.ip_address, "localhost")
sftp = paramiko.SFTPClient.from_transport(transport)
experiment_assets = [os.path.join(root, name) for root, dirs, files in os.walk("./templates/gstar") for name in files]
for expasset in experiment_assets:
sftp.put(expasset,expasset.replace("./templates","/root"))
sftp.close()
#install terminal browser w3m
stdin, stdout, stderr = ssh.exec_command("apt-get install -y w3m w3m-img htop")
else:
# Make a /etc/hosts file as you go
hosts.write(node.ip_address + "\t" + host_template + str(i) +"
")
# Set hostname on the machine
stdin, stdout, stderr = ssh.exec_command('echo \"'+host_template+str(i)+"\" > /etc/hostname")
stdin, stdout, stderr = ssh.exec_command('hostname \"'+host_template+str(i)+"\"")
ssh.exec_command('echo "JAVA_OPTS=\"-Xms'+str(int(int(self.utils.ram)*0.9))+'m -Xmx'+str(int(int(self.utils.ram)*0.9))+'m\"" >> /root/.bashrc')
ssh.exec_command('source /root/.bashrc')
# Add node to cluster
self.cluster[host_template+ str(i)] = node
ssh.close()
time.sleep(1)
# Save all collected known keys
ssh.save_host_keys("/tmp/known_hosts_"+str(i))
# Increase i
i = i+1
# Copy /etc/hosts on all clients
# Decrase to have the last node in i
i = i-1
hosts.close()
key_template_path="./templates/ssh_keys"
## Copy standard templates and name each node accordingly
for node in nodes:
ssh = paramiko.SSHClient()
ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
ssh.connect(node.ip_address, username='root', password='secretpw')
## Enlarge the user limit on open file descriptors
## (workaround for HDFS-127:http://wiki.apache.org/hadoop/Hbase/Troubleshooting#A7)
stdin, stdout, stderr = ssh.exec_command('ulimit -HSn 32768')
## Sync clocks
stdin, stdout, stderr = ssh.exec_command('service ntp stop')
stdin, stdout, stderr = ssh.exec_command('ntpd -gq')
stdin, stdout, stderr = ssh.exec_command('service ntp start')
transport = paramiko.Transport((node.ip_address, 22))
transport.connect(username = 'root', password = 'secretpw')
transport.open_channel("session", node.ip_address, "localhost")
sftp = paramiko.SFTPClient.from_transport(transport)
# Copy private and public key
sftp.put( key_template_path+"/id_rsa","/root/.ssh/id_rsa")
sftp.put( key_template_path+"/id_rsa.pub", "/root/.ssh/id_rsa.pub")
sftp.put( key_template_path+"/config", "/root/.ssh/config")
## Change permissions for private key
stdin, stdout, stderr = ssh.exec_command('chmod 0600 /root/.ssh/id_rsa')
# Add public key to authorized_keys
stdin, stdout, stderr = ssh.exec_command('cat /root/.ssh/id_rsa.pub >> /root/.ssh/authorized_keys')
with open('./templates/ssh_keys/authorized_keys') as f:
stdin, stdout, stderr = ssh.exec_command('echo \''+f.readline().strip()+'\' >> /root/.ssh/authorized_keys')
# print stdout.readlines()
# Copy files (/etc/hosts, masters, slaves and conf templates) removing empty lines
sftp.put( "/tmp/hosts", "/etc/hosts")
sftp.close()
ssh.close()
self.host_template = host_template
## Manipulate known hosts to make a good file
known_hosts_name = '/tmp/known_hosts'
known_hosts = open(known_hosts_name, 'w')
j = 0
while j <= i:
loop = open('/tmp/known_hosts_'+str(j), 'r')
for fileLine in loop.readlines():
known_hosts.write(fileLine.strip())
loop.close()
os.system("sed -i '/^$/d' /tmp/known_hosts_"+str(j))
j = j + 1
known_hosts.close()
for (clusterkey, clusternode) in self.cluster.items():
for line in fileinput.FileInput(known_hosts_name,inplace=1):
line = line.replace(clusternode.ip_address, clusterkey).strip()
print line
# print clusterkey, clusternode.public_dns_name
## Upload perfect file
for node in nodes:
transport = paramiko.Transport((node.ip_address, 22))
transport.connect(username = 'root', password = 'secretpw')
transport.open_channel("session", node.ip_address, "localhost")
sftp = paramiko.SFTPClient.from_transport(transport)
os.system("sed -i '/^$/d' /tmp/known_hosts")
sftp.put( "/tmp/known_hosts", "/root/.ssh/known_hosts")
sftp.close()
## Save to database
self.utils.add_to_cluster_db(self.cluster, self.cluster_id)
con = sqlite.connect(self.utils.db_file)
cur = con.cursor()
## Now you should be ok, so return the nodes with hostnames
return self.cluster
# def block_until_ping(self, nodes):
# ''' Blocks until all defined instances have reached running state and an ip has been assigned'''
# ## Run describe instances until everyone is running
# tmpnodes = nodes
# nodes = []
# while len(tmpnodes) > 0 :
# print "Waiting for", len(tmpnodes), "nodes."
# response = os.system("ping -c 1 -W 4 " + tmpnodes[0].private_dns_name)
# if response == 0:
# nodes.append(tmpnodes.pop(0));
# else:
# time.sleep(1);
# return nodes
def start_cluster (self):
print "nothing to start"
def stop_cluster (self):
print "nothing to stop"
|