集群监控工具简介：Ganglia和Nagios – 51CTO.COM

deathstar001</p><p>deathstar002</p><p>...skip a few...</p><p>deathstar099</p><p>deathstar100</p><p>

# for i in `cat /tmp/mynodes`; do </p><p>scp /usr/sbin/gmond $i:/usr/sbin/gmond</p><p>ssh $i mkdir -p /etc/ganglia/</p><p>scp /etc/ganglia/gmond.conf $i:/etc/ganglia/</p><p>scp /etc/init.d/gmond $i:/etc/init.d/</p><p>scp /usr/lib64/libganglia-3.1.1.so.0 $i:/usr/lib64/</p><p>scp /lib64/libexpat.so.0 $i:/lib64/</p><p>scp /usr/lib64/libconfuse.so.0 $i:/usr/lib64/</p><p>scp /usr/lib64/libapr-1.so.0 $i:/usr/lib64/</p><p>scp -r /usr/lib64/ganglia $i:/usr/lib64/</p><p>ssh $i service gmond start</p><p>done</p><p>

modules {</p><p>    module {</p><p>           name = "core_metrics"</p><p>     }</p><p>     ...</p><p>}</p><p>

  module {</p><p>    name = "python_module"</p><p>    path = "modpython.so"</p><p>    params = "/usr/lib64/ganglia/python_modules/"</p><p>  }</p><p>

mkdir /etc/ganglia/conf.d</p><p>mkdir /usr/lib64/ganglia/python_modules</p><p>

modprobe ipmi_msghandler</p><p>modprobe ipmi_si</p><p>modprobe ipmi_devintf</p><p>

Ambient_Temp,20,degrees_C,ok</p><p>CPU_1_Temp,20,degrees_C,ok</p><p>CPU_2_Temp,21,degrees_C,ok</p><p>

</p><p>import os</p><p>def temp_handler(name):</p><p>  # our commands we're going to execute</p><p>  sdrfile = "/tmp/sdr.dump"</p><p>  ipmitool = "/usr/bin/ipmitool"</p><p>  # Before you run this Load the IPMI drivers:</p><p>  # modprobe ipmi_msghandler</p><p>  # modprobe ipmi_si</p><p>  # modprobe ipmi_devintf</p><p>  # you'll also need to change permissions of /dev/ipmi0 for nobody</p><p>  # chown nobody:nobody /dev/ipmi0</p><p>  # put the above in /etc/rc.d/rc.local</p><p>  foo = os.path.exists(sdrfile)</p><p>  if os.path.exists(sdrfile) != True:</p><p>    os.system(ipmitool + ' sdr dump ' + sdrfile)</p><p>  if os.path.exists(sdrfile):</p><p>    ipmicmd = ipmitool + " -S " + sdrfile + " -c sdr"</p><p>  else:</p><p>    print "file does not exist... oops!"</p><p>    ipmicmd = ipmitool + " -c sdr"</p><p>  cmd = ipmicmd + " type temperature | sed 's/ /_/g' "</p><p>  cmd = cmd + " | awk -F, '/Ambient/ {print $2}' "</p><p>  #print cmd</p><p>  entries = os.popen(cmd)</p><p>  for l in entries:</p><p>    line = l.split()</p><p>  # print line</p><p>  return int(line[0])</p><p>def metric_init(params):</p><p>    global descriptors</p><p>    temp = {'name': 'Ambient Temp',</p><p>        'call_back': temp_handler,</p><p>        'time_max': 90,</p><p>        'value_type': 'uint',</p><p>        'units': 'C',</p><p>        'slope': 'both',</p><p>        'format': '%u',</p><p>        'description': 'Ambient Temperature of host through IPMI',</p><p>        'groups': 'IPMI In Band'}</p><p>    descriptors = [temp]</p><p>    return descriptors</p><p>def metric_cleanup():</p><p>    '''Clean up the metric module.'''</p><p>    pass</p><p>#This code is for debugging and unit testing</p><p>if __name__ == '__main__':</p><p>    metric_init(None)</p><p>    for d in descriptors:</p><p>        v = d['call_back'](d['name'])</p><p>        print 'value for %s is %u' % (d['name'],  v)</p><p>

</p><p>modules {</p><p>  module {</p><p>    name = "Ambient Temp"</p><p>    language = "python"</p><p>  }</p><p>}</p><p>collection_group {</p><p>  collect_every = 10</p><p>  time_threshold = 50</p><p>  metric {</p><p>    name = "Ambient Temp"</p><p>    title = "Ambient Temperature"</p><p>    value_threshold = 70</p><p>  }</p><p>}</p><p>

# ipmitool -I lanplus -H x01-bmc -U USERID -P PASSW0RD sdr dump \ /tmp/x01.sdr</p><p>Dumping Sensor Data Repository to '/tmp/x01.sdr'</p><p># ipmitool -I lanplus -H x01-bmc -U USERID -P PASSW0RD -S /tmp/x01.sdr \ sdr type </p><p>                                                                            Temperature</p><p>Ambient Temp     | 32h | ok  | 12.1 | 20 degrees C</p><p>CPU 1 Temp       | 98h | ok  |  3.1 | 20 degrees C</p><p>CPU 2 Temp       | 99h | ok  |  3.2 | 21 degrees C</p><p>

#!/usr/bin/perl</p><p># vallard@us.ibm.com</p><p>use strict;  # to keep things clean... er cleaner</p><p>use Socket;  # to resolve host names into IP addresses</p><p># code to clean up after forks</p><p>use POSIX ":sys_wait_h";</p><p># nodeFile: is just a plain text file with a list of nodes:</p><p># e.g:</p><p># node01</p><p># node02</p><p># ...</p><p># nodexx</p><p>my $nodeFile = "/usr/local/bin/nodes";</p><p># gmetric binary</p><p>my $gmetric = "/usr/bin/gmetric";</p><p>#ipmitool binary</p><p>my $ipmi = "/usr/bin/ipmitool";</p><p># userid for BMCs</p><p>my $u = "xcat";</p><p># password for BMCs</p><p>my $p = "f00bar";</p><p># open the nodes file and iterate through each node</p><p>open(FH, "$nodeFile") or die "can't open $nodeFile";</p><p>while(my $node = &lt;FH&gt;){</p><p>  # fork so each remote data call is done in parallel</p><p>  if(my $pid = fork()){</p><p>    # parent process</p><p>    next;</p><p>  }</p><p>  # child process begins here</p><p>  chomp($node);  # get rid of new line</p><p>  # resolve node's IP address for spoofing</p><p>  my $ip;</p><p>  my $pip = gethostbyname($node);</p><p>  if(defined $pip){</p><p>    $ip = inet_ntoa($pip);</p><p>  }else{</p><p>    print "Can't get IP for $node!\n";</p><p>    exit 1;</p><p>  }</p><p>  # check if the SDR cache file exists.</p><p>  my $ipmiCmd;</p><p>  unless(-f "/tmp/$node.sdr"){</p><p>    # no SDR cache, so try to create it...</p><p>    $ipmiCmd = "$ipmi -I lan -H $node-bmc -U $u -P $p sdr dump /tmp/$node.sdr";</p><p>    `$ipmiCmd`;</p><p>  }</p><p>  if(-f "/tmp/$node.sdr"){</p><p>    # run the command against the cache so that its faster</p><p>    $ipmiCmd = "$ipmi -I lan -H $node-bmc -U $u -P $p -S /tmp/$node.sdr sdr type </p><p>                                                                       Temperature ";</p><p>    # put all the output into the @out array</p><p>    my @out = `$ipmiCmd`;</p><p>    # iterate through each @out entry.</p><p>    foreach(@out){</p><p>      # each output line looks like this:</p><p>      # Ambient Temp     | 32h | ok  | 12.1 | 25 degrees C</p><p>      # so we parse it out</p><p>      chomp(); # get rid of the new line</p><p>      # grap the first and 5th fields.  (Description and Temp)</p><p>      my ($descr, undef, undef, undef,$temp) = split(/\|/);</p><p>      # get rid of white space in description</p><p>      $descr =~ s/ //g;</p><p>      # grap just the temp, (We assume C anyway)</p><p>      $temp = (split(' ', $temp))[0];</p><p>      # make sure that temperature is a number:</p><p>      if($temp =~ /^\d+/ ){</p><p>        #print "$node: $descr $temp\n";</p><p>        my $gcmd = "$gmetric -n '$descr' -v $temp -t int16 -u Celcius -S $ip:$node";</p><p>        `$gcmd`;</p><p>      }</p><p>    }</p><p>  }</p><p>  # Child Thread done and exits.</p><p>  exit;</p><p>}</p><p># wait for all forks to end...</p><p>while(waitpid(-1,WNOHANG) != -1){</p><p>  1;</p><p>}</p><p>