最近機房剛上了一批機器(有 100 台左右)，需要使用 Nagios 對這一批機器進行監控。領導要求兩天時間完成所有主機的監控。從原來的經驗來看，兩天時間肯定完成不了。那怎麼辦？按照之前的想法，肯定是在 nagios 配置文件逐一添加每台客戶端的監控信息，工作量巨大。突然，想到一個想法，是否可以通過腳本來實現批量對主機進行監控，也就是運維自動化。
寫腳本，最重要的就是思路。思路壓倒一切，經過思考最終決定就這麼做了。先貼出來一張網路拓撲圖：
圖片 9.1 pic
整個過程可以分為三部分。
- cmdb 端:主要用來實現對數據的收集，採用兩個 API，一個是提供給客戶機的 API。用於將客戶端的數據上傳的 cmdb 服務器；另外一 個 API 是 nagios 通過此 API 可以得到要監控主機的信息，然後對該信息進行整理，做成 nagios 監控模板。
- Client 端：通過 Python 腳本收集本機器要監控的軟硬件信息，然後通過 cmdb 端提供的 API 接口將數據上傳到 cmdb 端的數據庫。
- Nagios 端：通過 cmdb 端提供的 API 接口實現對 cmdb 收集到的信息進行抓取，然後將數據寫入到模板，最後 copy 到 naigos 指定的 objects 目錄，最終實現對客戶機的監控。
這三部分最重要的應該是 CMDB 端。接下來通過安裝 django 和編寫 API 接口實現 cmdb 可以正常工作。可以將 cmdb 端分為三個步驟來完成：
- 安裝 django
- 配置 django
- 編寫 API 接口
首先來進行安裝 django：
在安裝 django 之前首先應該安裝 python (版本建議 2.7.)
1.下載 django 軟件包 可以到 django 官方網站下載最新 django 軟件包(https://www.djangoproject.com).2.解壓縮並安裝軟件包 tar -zxvf Django-1.5.1.tar.gz cd Django-1.5.1 python setup.py install
創建項目和應用：
1.創建一個項目python startproject simplecmdb2.創建一個應用python startapp hostinfo
配置 django：
1.修改 setting.py
DATABASES = {\'ENGIN\':\'django.db.backends.sqlite\',\'name\':path.join(\'CMDB.db\')} #使用的數據庫及數據庫名INSTALLED_APPS =（hostinfoINSTALLED_APPS = （\'hostinfo\'）INSTALLED_APPS = (\'hostinfo\') #應用的名稱 2.修改urls.py
url(r\'^api/gethost.json$\',\'hostinfo.views.gethosts\'), #nagios客戶端訪問API接口地址url(r\'^api/clooect$\',\'hostinfo.views.collect\'), #客戶端訪問API進行上傳數據的APIurl(r\'^admin/\',include(admin.site.urls)), #django後台管理登入urlfrom django.contrib import adminadmin.autodiscover
3.修改項目 hostinfo 下的 views.py
代碼如下：
# Create your views here. #包含以下模塊from django.shortcuts import render_to_responsefrom django.http import HttpResponsefrom models import Host, HostGroup #包含json模塊try: import jsonexcept ImportError,e: import simplejson as json #用來接收客戶端服務器發送過來的數據def collect(request): req = request if req.POST:vendor = req.POST.get(\'Product_Name\')sn = req.POST.get(\'Serial_Number\')product = req.POST.get(\'Manufacturer\')cpu_model = req.POST.get(\'Model_Name\')cpu_num = req.POST.get(\'Cpu_Cores\')cpu_vendor = req.POST.get(\'Vendor_Id\')memory_part_number = req.POST.get(\'Part_Number\')memory_manufacturer = req.POST.get(\'Manufacturer\')memory_size = req.POST.get(\'Size\')device_model = req.POST.get(\'Device_Model\')device_version = req.POST.get(\'Firmware_Version\')device_sn = req.POST.get(\'Serial_Number\')device_size = req.POST.get(\'User_Capacity\')osver = req.POST.get(\'os_version\')hostname = req.POST.get(\'os_name\')os_release = req.POST.get(\'os_release\')ipaddrs = req.POST.get(\'Ipaddr\')mac = req.POST.get(\'Device\')link = req.POST.get(\'Link\')mask = req.POST.get(\'Mask\')device = req.POST.get(\'Device\')host = Hosthost.hostname = hostnamehost.product = producthost.cpu_num = cpu_numhost.cpu_model = cpu_modelhost.cpu_vendor = cpu_vendorhost.memory_part_number = memory_part_numberhost.memory_manufacturer = memory_manufacturerhost.memory_size = memory_sizehost.device_model = device_modelhost.device_version = device_versionhost.device_sn = device_snhost.device_size = device_sizehost.osver = osverhost.os_release = os_releasehost.vendor = vendorhost.sn = snhost.ipaddr = ipaddrshost.save #將客戶端傳過來的數據通過POST接收，存入數據庫return HttpResponse(\'OK\') #如果插入成功，返回\'ok\' else:return HttpResponse(\'no post data\') #提供給NAGIOS 的APIdef gethosts(req): d = hostgroups = HostGroup.objects.all for hg in hostgroups:ret_hg = {\'hostgroup\':hg.name,\'members\':}members = hg.members.allfor h in members: ret_h = {\'hostname\':h.hostname, #API接口返回的數據\'ipaddr\':h.ipaddr } ret_hg[\'members\'].append(ret_h)d.append(ret_hg) ret = {\'status\':0,\'data\':d,\'message\':\'ok\'} return HttpResponse(json.dumps(ret)) 4.修改 model.py 文件
代碼如下：
from django.db import models # Create your models here. #插入數據庫的Host表,主要存儲客戶端主機的信息class Host(models.Model): """store host information""" vendor = models.CharField(max_length=30,null=True) sn = models.CharField(max_length=30,null=True) product = models.CharField(max_length=30,null=True) cpu_model = models.CharField(max_length=50,null=True) cpu_num = models.CharField(max_length=2,null=True) cpu_vendor = models.CharField(max_length=30,null=True) memory_part_number = models.CharField(max_length=30,null=True) memory_manufacturer = models.CharField(max_length=30,null=True) memory_size = models.CharField(max_length=20,null=True) device_model = models.CharField(max_length=30,null=True) device_version = models.CharField(max_length=30,null=True) device_sn = models.CharField(max_length=30,null=True) device_size = models.CharField(max_length=30,null=True) osver = models.CharField(max_length=30,null=True) hostname = models.CharField(max_length=30,null=True) os_release = models.CharField(max_length=30,null=True) ipaddr = models.IPAddressField(max_length=15) def __unicode__(self):return self.hostname #主機組表，用來對主機進行分組class HostGroup(models.Model): name = models.CharField(max_length=30) members = models.ManyToManyField(Host)
5.修改 admin.py 文件
#from models import Host, IPaddrfrom models import Host, HostGroupfrom django.contrib import admin #設置在django在admin後天顯示的名稱class HostAdmin(admin.ModelAdmin): list_display = [\'vendor\',\'sn\',\'product\',\'cpu_model\',\'cpu_num\',\'cpu_vendor\',\'memory_part_number\',\'memory_manufacturer\',\'memory_size\',\'device_model\',\'device_version\',\'device_sn\',\'device_size\',\'osver\',\'hostname\',\'os_release\'] #在django後台amdin顯示的組名稱class HostGroupAdmin(admin.ModelAdmin): list_display = [\'name\',] #將如上兩個類的數據展示到django的後台admin.site.register(HostGroup,HostGroupAdmin)admin.site.register(Host, HostAdmin)
6.創建數據庫
python manager.py syncdb #創建數據庫
7.啟動應用
python manager.py runserver 0.0.0.0:8000
8.測試
http://132.96.77.12:8000/admin
圖片 9.2 pic
圖片 9.3 pic
通過上圖可以看到，django 已經配置成功。
接下來可以在客戶端編寫收集主機信息的腳本了，主要抓取 cpu、內存、硬盤、服務器型號、服務器 sn、ip 地址、主機名稱、操作系統版本等信息，共 7 個腳本:
1.cpu 抓取腳本：
#!/usr/local/src/python/bin/python #-*- coding:utf-8 -*-from subprocess import PIPE,Popenimport redef getCpuInfo: p = Popen([\'cat\',\'/proc/cpuinfo\'],shell=False,stdout=PIPE) stdout, stderr = p.communicate return stdout.stripdef parserCpuInfo(cpudata): pd = {} model_name = re.compile(r\'.*model names+:s(.*)\') vendor_id = re.compile(r\'vendor_ids+:(.*)\') cpu_cores = re.compile(r\'cpu coress+:s([d]+)\') lines = [line for line in cpudata.split(\'n\')] for line in lines:model = re.match(model_name,line)vendor = re.match(vendor_id,line)cores = re.match(cpu_cores,line)if model: pd[\'Model_Name\'] = model.groups[0].stripif vendor: pd[\'Vendor_Id\'] = vendor.groups[0].stripif cores: pd[\'Cpu_Cores\'] = cores.groups[0]else: pd[\'Cpu_Cores\'] = int(\'1\') return pdif __name__ == \'__main__\': cpudata = getCpuInfo print parserCpuInfo(cpudata) 2.硬盤抓取腳本：
#!/usr/local/src/python/bin/python #-*- coding:utf-8 -*-from subprocess import PIPE,Popenimport redef getDiskInfo: disk_dev = re.compile(r\'Disks/dev/[a-z]{3}\') disk_name = re.compile(r\'/dev/[a-z]{3}\') p = Popen([\'fdisk\',\'-l\'],shell=False,stdout=PIPE) stdout, stderr = p.communicate for i in stdout.split(\'n\'):disk = re.match(disk_dev,i)if disk: dk = re.search(disk_name,disk.group).group n = Popen(\'smartctl -i %s\' % dk,shell=True,stdout=PIPE) stdout, stderr = n.communicate return stdout.stripdef parserDiskInfo(diskdata): ld = pd = {} device_model = re.compile(r\'(Device Model):(s+.*)\') serial_number = re.compile(r\'(Serial Number):(s+[dw]{1,30})\') firmware_version = re.compile(r\'(Firmware Version):(s+[w]{1,20})\') user_capacity = re.compile(r\'(User Capacity):(s+[dw, ]{1,50})\') for line in diskdata.split(\'n\'):serial = re.search(serial_number,line)device = re.search(device_model,line)firmware = re.search(firmware_version,line)user = re.search(user_capacity,line)if device: pd[\'Device_Model\'] = device.groups[1].stripif serial: pd[\'Serial_Number\'] = serial.groups[1].stripif firmware: pd[\'Firmware_Version\'] = firmware.groups[1].stripif user: pd[\'User_Capacity\'] = user.groups[1].strip return pdif __name__ == \'__main__\': diskdata = getDiskInfo print parserDiskInfo(diskdata) 3.內存抓取腳本：
#!/usr/local/src/python/bin/python #-*- coding:utf-8 -*-from subprocess import PIPE,Popenimport reimport sysdef getMemInfo: p = Popen([\'dmidecode\'],shell=False,stdout=PIPE) stdout, stderr = p.communicate return stdout.stripdef parserMemInfo(memdata): line_in = False mem_str = \'\' pd = {} fd = {} for line in memdata.split(\'n\'):if line.startswith(\'Memory Device\') and line.endswith(\'Memory Device\'): line_in = True mem_str+=\'n\' continueif line.startswith(\'t\') and line_in: mem_str+=lineelse: line_in = False for i in mem_str.split(\'n\')[1:]:lines = i.replace(\'t\',\'n\').stripfor ln in lines.split(\'n\'): k, v = [i for i in ln.split(\':\')] pd[k.strip] = v.stripif pd[\'Size\'] != \'No Module Installed\': mem_info = \'Size:%s ; Part_Number:%s ; Manufacturer:%s\' % (pd[\'Size\'],pd[\'Part Number\'],pd[\'Manufacturer\']) for line in mem_info.split(\'n\'):for word in line.split(\';\'): k, v = [i.strip for i in word.split(\':\')] fd[k] = v.stripyield fdif __name__ == \'__main__\': memdata = getMemInfo for i in parserMemInfo(memdata):print i 4.抓取服務器信息腳本：
#!/usr/local/src/python/bin/python # -*- coding:utf-8 -*-from subprocess import PIPE,Popenimport urllib, urllib2def getDMI: p = Popen(\'dmidecode\',shell=True,stdout=PIPE) stdout, stderr = p.communicate return stdoutdef parserDMI(dmidata): pd = {} fd = {} line_in = False for line in dmidata.split(\'n\'):if line.startswith(\'System Information\'): line_in = True continueif line.startswith(\'t\') and line_in: k, v = [i.strip for i in line.split(\':\')] pd[k] = velse: line_in = False name = "Manufacturer:%s ; Serial_Number:%s ; Product_Name:%s" % (pd[\'Manufacturer\'],pd[\'Serial Number\'],pd[\'Product Name\']) for i in name.split(\';\'):k, v = [j.strip for j in i.split(\':\')]fd[k] = v return fdif __name__ == \'__main__\': dmidata = getDMI postdata = parserDMI(dmidata) print postdata 5.抓取主機信息
#!/usr/local/src/python/bin/python #-*- coding:utf-8 -*-import platformdef getHostInfo: pd ={} version = platform.dist os_name = platform.node os_release = platform.release os_version = \'%s %s\' % (version[0],version[1]) pd[\'os_name\'] = os_name pd[\'os_release\'] = os_release pd[\'os_version\'] = os_version return pdif __name__ == \'__main__\': print getHostInfo 6.抓取 ip 地址：
#!/usr/local/src/python/bin/python #-*- coding:utf-8 -*-from subprocess import PIPE,Popenimport redef getIpaddr: p = Popen([\'ifconfig\'],shell=False,stdout=PIPE) stdout, stderr = p.communicate return stdout.stripdef parserIpaddr(ipdata): device = re.compile(r\'(ethd)\') ipaddr = re.compile(r\'(inet addr:[d.]{7,15})\') mac = re.compile(r\'(HWaddrs[0-9A-Fa-f:]{17})\') link = re.compile(r\'(Link encap:[w]{3,14})\') mask = re.compile(r\'(Mask:[d.]{9,15})\') for lines in ipdata.split(\'nn\'):pd = {}eth_device = re.search(device,lines)inet_ip = re.search(ipaddr,lines)hw = re.search(mac,lines)link_encap = re.search(link,lines)_mask = re.search(mask,lines)if eth_device: if eth_device:Device = eth_device.groups[0] if inet_ip:Ipaddr = inet_ip.groups[0].split(\':\')[1] if hw:Mac = hw.groups[0].split[1] if link_encap:Link = link_encap.groups[0].split(\':\')[1] if _mask:Mask = _mask.groups[0].split(\':\')[1] pd[\'Device\'] = Device pd[\'Ipaddr\'] = Ipaddr pd[\'Mac\'] = Mac pd[\'Link\'] = Link pd[\'Mask\'] = Mask yield pdif __name__ == \'__main__\': ipdata = getIpaddr for i in parserIpaddr(ipdata):print i 7.對這些信息進行合併，並通過API形式將數據發送給cmdb端
#!/usr/local/src/python/bin/pythonimport urllib, urllib2from cpuinfo import *from diskinfo import *from meminfo import *from product import *from hostinfo import *from ipaddress import *def getHostTotal: ld = cpuinfo = parserCpuInfo(getCpuInfo) diskinfo = parserDiskInfo(getDiskInfo) for i in parserMemInfo(getMemInfo):meminfo = i productinfo = parserDMI(getDMI) hostinfo = getHostInfo ipaddr = parserIpaddr(getIpaddr) for i in ipaddr:ip = i for k in cpuinfo.iteritems:ld.append(k) for i in diskinfo.iteritems:ld.append(i) for j in meminfo.iteritems:ld.append(j) for v in productinfo.iteritems:ld.append(v) for x in hostinfo.iteritems:ld.append(x) for y in ip.iteritems:ld.append(y) return lddef parserHostTotal(hostdata): pg = {} for i in hostdata:pg[i[0]] = i[1] return pgdef urlPost(postdata): data = urllib.urlencode(postdata) req = urllib2.Request(\'http://132.96.77.12:8000/api/collect\',data) response = urllib2.urlopen(req) return response.readif __name__ == \'__main__\': hostdata = getHostTotal postdata = parserHostTotal(hostdata) print urlPost(postdata) 到目前為止，cmdb 系統已經可以將所有客戶端的主機信息寫入到數據庫，並且可以通過 nagios 端的 API 接口直接調到數據：
http://132.96.77.12:8000/api/gethosts.json
圖片 9.4 pic
通過圖可以看到，已經成功調用到 API 接口的數據。
接下來可以在 nagios 端進行調用 API 接口的數據，對數據進行格式化。並寫入文件。
1.nagios腳本
#!/opt/data/py/bin/python #!-*- coding:utf-8 -*-import urllib, urllib2import jsonimport osimport shutilCURR_DIR = os.path.abspath(os.path.dirname(__file__))HOST_CONF_DIR = os.path.join(CURR_DIR,\'hosts\')HOST_TMP = """define host { uselinux-server host_name %(hostname)s check_command check-host-alive alias %(hostname)s address %(ipaddr)s contact_groups admins}"""def getHosts: url = \'http://132.96.77.12:8000/api/gethosts.json\' return json.loads(urllib2.urlopen(url).read)def initDir: if not os.path.exists(HOST_CONF_DIR):os.mkdir(HOST_CONF_DIR)def writeFile(f,s): with open(f,\'w\') as fd:fd.write(s)def genNagiosHost(hostdata): initDir conf = os.path.join(HOST_CONF_DIR,\'hosts.cfg\') hostconf = "" for hg in hostdata:for h in hg[\'members\']: hostconf+=HOST_TMP %h writeFile(conf,hostconf) return "ok"def main: result = getHosts if result[\'status\'] == 0:print genNagiosHost(result[\'data\']) else:print \'Error: %s\' % result[\'message\'] if os.path.exists(os.path.join(HOST_CONF_DIR,\'hosts.cfg\')):os.chdir(HOST_CONF_DIR)shutil.copyfile(\'hosts.cfg\',\'/etc/nagios/objects/hosts.cfg\')if __name__ == "__main__": main 現在已經生成 nagios 主機的配置文件，並 copy 到 nagios/objects 目錄下 hosts.cfg。接下來可以測試是否 nagios 配置有問題，如果沒有問題，就可以啟動 nagios 服務
[root@yetcomm-v2 bin]# ./nagios -v /etc/nagios/nagios.cfg
通過測試，nagios 沒有發生錯誤或警告信息，現在可以啟動 nagios 服務：
[root@yetcomm-v2 bin]# service nagios restart
最後，可以通過瀏覽器查看 nagios 的監控界面：
圖片 9.5 pic
通過上圖，可以看到已經將一台主機加入到監控組。由於是生產環境，所有只能拿測試服務器進行測試。其實測試環境和生產環境的代碼完全一致。