def aboutCross(self):
## 检查跨服连接状态, 1, 异常; 0, 正常;
## 750 是组队竞技场，战场，帮派战，751 是组队副本，襄阳城和跨服运镖
try:
dbName = self.srvFlag + "_dbtx"
conn = MySQLdb.connect(host=dbOneIp,port=db1Port,user=dbUser,passwd=dbPass,db=dbName,charset=dbChar)
cursor = conn.cursor(cursorclass = MySQLdb.cursors.DictCursor)
dSql = "select public_url from t_cross_copy_channel where channel=750 limit 1;"
ySql = "select public_url from t_cross_copy_channel where channel=751 limit 1;"
cursor.execute(dSql)
pkRes = cursor.fetchone()
## pkUrl = cursor.fetchone()['public_url']
## pkDomain, pkPort = pkUrl.split(":")
cursor.execute(ySql)
fbRes = cursor.fetchone()
## fbUrl = cursor.fetchone()['public_url']
## fbDomain, fbPort = fbUrl.split(":")
crossSwStat, fixSwStat = self.sDb.read(self.srvFlag, "crossAlarm"), self.sDb.read(self.srvFlag, "isFixing")
try:
pkUrl, fbUrl = pkRes['public_url'], fbRes['public_url']
except Exception, err:
saveLog.warning("get Cross Domain error occured. %s" % str(err))
crossStatus, pkDomain, fbDomain = 1, 'null', 'null'
## if pkUrl and fbUrl:
else:
def getSrvVer(self, srvPidStatus):
## 服务端版本号
try:
DbAppPid = int(srvPidStatus['DbCacheApp_' + self.srvFlag]['DbCacheApp_' + self.srvFlag + "_pid"])
except Exception,err:
srvVersion = "null"
saveLog.warning("Failed to get DbCacheApp's PID, %s" % str(err))
def getDisk(self):
mountPoints, diskTotal = [], 0 #diskTotal累加
parts = psutil.disk_partitions() #分区信息
for p in parts:
mp = str(p).split(",")[1].split("=")[1]
mountPoints.append(mp)
for x in mountPoints:
pTotal = psutil.disk_usage(x.strip("'")).total #单个分区容量
diskTotal += pTotal
diskTotal = changeUnit(int(diskTotal))
rootSize, rootRate = psutil.disk_usage('/').free, psutil.disk_usage('/').percent #获取/分区的剩余空间和使用率
writeToFile("totalDisk", diskTotal)
writeToFile("rootSize", changeUnit(int(rootSize)))
writeToFile("rootRate", rootRate)
if float(rootRate) > float(85):
"""ON开关，防止多次重复报警；报警一次之后设为关闭"""
if self.sDb.read(self.telIP, "rootAlarm") == "ON": ## 状态为"NO"且开关为"ON"时,报故障
sub = "DBTX ROOT partition Alarm:%s" % self.ps1
msg = "DateTime: %s\n根分区已使用%s%%, 超过85%%, 请警惕!!!" % (getTimeNow(), rootRate)
saveLog.warning(msg)
sendMail(sub, msg)
self.sDb.update(self.telIP, "rootAlarm", "OFF") ## 报警后,将开关置为"OFF"
else:
if self.sDb.read(self.telIP, "rootAlarm") == "OFF": ## 状态OK且开关为"OFF"时,报恢复
sub = "DBTX ROOT partition is OK:%s" % self.ps1
msg = "DateTime: %s\n根分区可用空间充足%s." % (getTimeNow(), changeUnit(int(rootSize)))
saveLog.info(msg)
sendMail(sub, msg)
self.sDb.update(self.telIP, "rootAlarm", "ON") ## 使用率不足85%时,将开关置为"ON"
def chkPort(tAddr, tPort): ## 检查端口号
import saveLog
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.settimeout(5)
try:
s.connect((tAddr, int(tPort)))
return True
except Exception,err:
saveLog.warning("chkPort() error occured. %s" % str(err))
return False
def sigRmAct(self, fPath): ## 删除单个文件/目录
ap = os.path.abspath(fPath)
splPath = ap.split("/")
if os.path.ismount(fPath): ## 判断是否是挂载点,挂载点严禁删除
saveLog.error("This path is a mount point,can't rm !!!")
sys.exit(1)
elif os.path.isdir(fPath): ## 如果是目录
if splPath[1] == "" or splPath[1] == "n*":
saveLog.error("You may deleting root-dir,can't rm !!!")
sys.exit(2)
elif splPath[1] in self.sysDirList:
saveLog.error("This path is in system-dir-lists, can't rm !!!")
sys.exit(2)
elif splPath[1] in self.riskDirList:
saveLog.warning("This path is in risk_dir_lists, move to %s... " % self.tmpDir)
refPath = ap.replace("/","-") ## 重命名,将绝对路径中的"/"替换成"-",以保存完整路径
oTime = str(datetime.datetime.now()).replace(" ","_")
shutil.move(ap, self.tmpDir+"/"+refPath+"_"+oTime)
else:
saveLog.info("Deleting %s... " % ap)
shutil.rmtree(ap)
elif os.path.isfile(fPath): ## 如果是文件
if splPath[1] in self.riskDirList:
if os.path.basename(ap) in self.riskFileList:
saveLog.error("Important document(file),can't rm !!!")
sys.exit(3)
saveLog.warning("This path(file) is in risk_dir_lists, move to %s..." % self.tmpDir)
refPath = ap.replace("/","-") ## 重命名,将绝对路径中的"/"替换成"-",以保存完整路径
oTime = str(datetime.datetime.now()).replace(" ","_")
shutil.move(ap, self.tmpDir+"/"+refPath+"_"+oTime)
elif splPath[1] in self.sysDirList:
saveLog.error("This path is in system-dir-lists, can't rm !!!")
sys.exit(2)
else:
saveLog.info("Deleting %s..." % ap)
os.remove(ap)
elif os.path.islink(fPath): ## 是链接的话,先找出它的真实路径
realPath = os.path.realpath(fPath)
splPath = realPath.split("/")
if splPath[1] in self.riskDirList:
if os.path.basename(ap) in self.riskFileList:
saveLog.error("Important document(link),can't rm !!!")
sys.exit(3)
saveLog.warning("This path(link) is in risk_dir_lists, move to %s..." % self.tmpDir)
refPath = ap.replace("/","-") ## 重命名,将绝对路径中的"/"替换成"-",以保存完整路径
oTime = str(datetime.datetime.now()).replace(" ","_")
shutil.move(ap, self.tmpDir+"/"+refPath+"_"+oTime)
elif splPath[1] in self.sysDirList:
saveLog.error("This path is in system-dir-lists, can't rm !!!")
sys.exit(2)
else:
saveLog.info("Deleting %s..." % ap)
os.remove(ap)
else:
pass
def getAppStatus(self):
## 检查游戏程序运行状态, 0, 关闭; 1, 正常; 2, 异常
srvFlag = str(self.srvFlag)
srvStatus, srvPidStatus, mUser = {'Success': [], 'Error': []}, {}, "*****@*****.**"
## srvNeedStart = ['DbCacheApp','LogDbApp','MailApp','SnapshotApp','CopyApp','CellMgrApp','CellApp','InterApp','GateMgrApp','GateApp','LoginApp']
for srv in gameApps:
try:
srv += "_" + srvFlag # App名_区服标识
srvPid = getProcessID(srv) # 调用shell pidof 获取pid
if "GateApp" in srv:
if srvPid:
GateAppCount = len(srvPid.split()) # GateApp个数
else:
GateAppCount = 0
if srvPid:
srvStatus['Success'].append(srv)
srvPidStatus[srv] = {srv+"_status": 1, srv+"_pid": srvPid}
else:
srvStatus['Error'].append(srv)
srvPidStatus[srv] = {srv+"_status": 0, srv+"_pid": None}
except Exception, err:
saveLog.warning("getAppStatus() error occured while get proce pid, %s" % str(err))
def gateAnalyze(srvFlag, nowNu, sDb):
from dbtx.Merge.etc import ps1
pubFile = "/opt/dbtx/" + str(srvFlag) + "/config/public_config.xml"
pubObj = open(pubFile)
fixGateNum = ''.join([ l.strip().split(">")[1].split("<")[0] for l in pubObj.readlines() if "<GateNum>" in l])
pubObj.close()
corMuser = "******"
if int(nowNu) < int(fixGateNum):
if sDb.read(srvFlag, "isFixing") == "OFF": ## 如果维护标识为"OFF"
if sDb.read(srvFlag, "appNumAlarm") == "ON": ## 状态为"NO"且开关为"ON"时,报故障
sub = "DBTX GateApps Alarm: %s" % ps1
msg = "DateTime: %s\n%s 预设开启 %s 个GateApp,当前只有 %s 个,挂了 %s 个 !!!" % (getTimeNow(), srvFlag, fixGateNum, nowNu, (int(fixGateNum)-int(nowNu)))
saveLog.warning(msg)
sendMail(sub, msg, corMuser)
sDb.update(srvFlag, "appNumAlarm", "OFF") ## 报警后,将开关置为"OFF"
else:
if sDb.read(srvFlag, "isFixing") == "OFF": ## 如果维护标识为"OFF"
if sDb.read(srvFlag, "appNumAlarm") == "OFF": ## 状态OK且开关为"OFF"时,报恢复
sub = "DBTX GateApps is OK: %s" % ps1
msg = "DateTime: %s\n%s GateApp Numbers正常." % (getTimeNow(), srvFlag)
saveLog.warning(msg)
sendMail(sub, msg, corMuser)
sDb.update(srvFlag, "appNumAlarm", "ON")
def appCpuAnalyze(srvPidStatus, srvFlag, sDb):
from dbtx.Merge.etc import ps1
corMuser = "******"
tNowCpu = topNow()
if not sDb.exists(srvFlag, "ct"): ## Redis库中不存在该服区标识,初始化游戏程序占用CPU达到95%的次数
sDb.intAppCpuCt(srvFlag)
for app in srvPidStatus:
appName = app.split("_")[0]
if srvPidStatus[app][app+"_status"] == 0 or "Snapshot" in app: ## 如果app状态异常,不检测CPU,MEM信息,忽略掉Snapshot
continue
appPid = srvPidStatus[app][app+"_pid"].split()
for pid in appPid: ## CellApp,GateApp,CopyApp会有多个进程ID
nCpu, appCount = tNowCpu[pid], int(sDb.read(appName, srvFlag))
if float(nCpu) > 95.5: ## CPU使用率都大于95%时,count+1
appCount += 1
sDb.update(appName, srvFlag, appCount) ## 将新的count值更新到Redis库
else:
sDb.update(appName, srvFlag, 0) ## 如果没有超过95%,则置为0
if int(appCount) >= 8:
if sDb.read(srvFlag, "isFixing") == "OFF": ## 如果维护标识为"OFF"
if sDb.read(srvFlag, "appCpuAlarm") == "ON": ## 状态为"NO"且开关为"ON"时,报故障
sub = "DBTX App-CPU Alarm: %s" % ps1
msg = "DateTime: %s\n%s CPU使用率连续五次达到95%%, 进程ID为: %s !!!" % (getTimeNow(), app, pid)
saveLog.warning(msg)
sendMail(sub, msg, corMuser)
## sDb.updateDb(mulDb, srvFlag, "appCpuAlarm", "OFF") ## 报警后,将开关置为"OFF"
sDb.update(appName, srvFlag, 0) ## 报警后,重置count值
if "DbCacheApp_"+srvFlag == app:
proc = psutil.Process(int(pid))
shr = proc.get_ext_memory_info()[2]
if float(shr) > 1932735283.2: ## DbCacheApp共享内存大于1.8G时,报警
if sDb.read(srvFlag, "isFixing") == "OFF": ## 如果维护标识为"OFF"
if sDb.read(srvFlag, "appShrAlarm") == "ON": ## 状态为"NO"且开关为"ON"时,报故障
sub = "DBTX DbCacheApp-SHR Alarm: %s" % ps1
msg = "DateTime: %s\n%s DbCacheApp 共享内存达到红色警戒,当前占用共享内存: %s !!!" % (getTimeNow(), srvFlag, changeUnit(shr))
saveLog.warning(msg)
sendMail(sub, msg, corMuser)
sDb.update(srvFlag, "appShrAlarm", "OFF") ## 报警后,将开关置为"OFF"
else:
if sDb.read(srvFlag, "isFixing") == "OFF": ## 如果维护标识为"OFF"
if sDb.read(srvFlag, "appShrAlarm") == "OFF": ## 状态OK且开关为"OFF"时,报恢复
sub = "DBTX DbCacheApp-SHR Alarm: %s" % ps1
msg = "DateTime: %s\n%s DbCacheApp 占用共享内存恢复到正常水平: %s ." % (getTimeNow(), srvFlag, changeUnit(shr))
saveLog.warning(msg)
sendMail(sub, msg, corMuser)
sDb.update(srvFlag, "appShrAlarm", "ON")
def getNginxStatus(self):
try:
if os.path.exists("/usr/local/nginx/logs/nginx.pid"):
fObj = open("/usr/local/nginx/logs/nginx.pid")
ngxMainPid = fObj.read().strip()
fObj.close()
else:
ngxMainPid = 0
pids = getProcessID("nginx")
if ngxMainPid:
if not pids or len(pids) < 8:
wp = ""
res = os.popen("ps -C nginx -o pid,cmd").readlines()
for l in res:
wp += "\n" + l.strip()
if self.sDb.read(telIP, "ngxAlarm") == "ON": ## 状态为"NO"且开关为"ON"时,报故障
sub = "DBTX Nginx Alarm: %s" % ps1
msg = "Datetime: %s\nNginx主进程PID:%s\nWorker_processes:%s\n少于10个!!" % (getTimeNow(), ngxMainPid, wp)
saveLog.warning(msg)
sendMail(sub, msg)
self.sDb.update(telIP, "ngxAlarm", "OFF")
else:
if self.sDb.read(telIP, "ngxAlarm") == "OFF": ## 状态OK且开关为"OFF"时,报恢复
sub = "DBTX Nginx is OK: %s" % ps1
msg = "Datetime: %s\nNginx主进程恢复开启状态." % (getTimeNow())
saveLog.info(msg)
sendMail(sub, msg)
self.sDb.update(telIP, "ngxAlarm", "ON")
else:
if self.sDb.read(telIP, "ngxAlarm") == "ON": ## 状态为"NO"且开关为"ON"时,报故障
sub = "DBTX Nginx Alarm: %s" % ps1
msg = "Datetime: %s\nNginx主进程挂了!!" % (getTimeNow())
saveLog.warning(msg)
sendMail(sub, msg)
self.sDb.update(telIP, "ngxAlarm", "OFF")
except Exception,err:
saveLog.warning("getNginxStatus, %s" % str(err))
redisData["command_rate"] = (int(redisData["total_commands"]) - int(last_commands)) / float(60)
fZbx = str(redisData).strip("{}").replace(",","\n").replace("'","") ## 转换成zabbix需要的格式
fObj = open(rZbx,'w')
fObj.write(fZbx)
fObj.close()
except Exception,err:
saveLog.error("getRedisBaseInfo(): %s" % str(err))
## 分析数据,必要时报警: 1, 使用内存大于1G; 2, 命中率低于20%
if int(redisData["used_memory"]) > 1073741824:
sDb = opRedis() ## 实例化 Redis库操作类
if sDb.read(telIP, "rRateAlarm") == "ON":
sub = "DBTX Redis Memory Alarm: %s" % ps1
msg = "DateTime: %s\nRedis使用内存为: %s, 超过1G,请警惕 !!!" % (getTimeNow(), changeUnit(int(redisData["used_memory"])))
saveLog.warning(msg)
sendMail(sub, msg)
sDb.update(telIP, "rRateAlarm", "OFF") ## 报警后,将开关置为"OFF"
else:
sDb = opRedis() ## 实例化 Redis库操作类
if sDb.read(telIP, "rRateAlarm") == "OFF":
sub = "DBTX Redis Memory is OK: %s" % ps1
msg = "DateTime: %s\nRedis使用内存情况正常: %s." % (getTimeNow(), changeUnit(int(redisData["used_memory"])))
saveLog.info(msg)
sendMail(sub, msg)
sDb.update(telIP, "rRateAlarm", "ON")
## if totalHits != 0 and redisData["hit_rate"] < 20:
## sDb = opShelveDb() ## 实例化 shelve库操作类
## if sDb.readDb(pubDb, telIP, "rRateAlarm") == "ON":
## sub = "DBTX Redis HitRate Alarm: %s" % ps1
def runCollect(self):
try:
from dbtx.Merge.etc import telIP,cncIP,lanIP,ps1,dbOneIp,dbTwoIp,cpuCores,totalDisk,rootSize,rootRate,net_adapter_type,net_adapter_driver_version,net_bandwidth,totalMem,role_type,ser_vendor,ser_product
except Exception, err:
saveLog.warning("Import Error. %s" % str(err))
os.system("/usr/bin/python /data/sh/monitor/Alive/doCollectFixInfo.py")
def getCoreFile(self, srvRootDir):
## 监控coredown
try:
corBaseDir = os.path.dirname(srvRootDir)
except Exception,err:
saveLog.warning("getCoreDir, %s" % str(err))
if appSwStat == "ON": ## 状态为"NO"且开关为"ON"时,报故障
sub = "DBTX AppService Alarm: %s" % ps1
msg = "DateTime: %s\n%s 挂啦 !!!" % (getTimeNow(), srvStatus['Error'])
saveLog.warning(msg)
sendMail(sub, msg, mUser)
self.sDb.update(srvFlag, "appAlarm", "OFF")
except Exception, err:
saveLog.warning("app_status is 2,get switch status error occured.")
elif srvStatus['Error'] and not srvStatus['Success']: ## 全部处于关闭状态
app_status = 0
try:
if fixSwStat == "OFF": ## 如果维护标识为"OFF"
if appSwStat == "ON": ## 状态为"NO"且开关为"ON"时,报故障
sub = "DBTX AppService is Down: %s" % ps1
msg = "DateTime: %s\n%s AppService 关闭 !!!" % (getTimeNow(), srvFlag)
saveLog.warning(msg)
sendMail(sub, msg, mUser)
self.sDb.update(srvFlag, "appAlarm", "OFF")
except Exception, err:
saveLog.warning("app_status is 0,get switch status error occured.")
elif not srvStatus['Error'] and srvStatus['Success']: ## 全部处于开启状态
app_status = 1
appCpuAnalyze(srvPidStatus,srvFlag,self.sDb)
self.getSrvVer(srvPidStatus)
gateAnalyze(srvFlag,GateAppCount,self.sDb)
try:
if fixSwStat == "OFF": ## 如果维护标识为"OFF"
if appSwStat == "OFF": ## 状态OK且开关为"OFF"时,报恢复
sub = "DBTX AppService is OK: %s" % ps1
msg = "DateTime: %s\n%s AppService 恢复开启 !!!" % (getTimeNow(), srvFlag)
saveLog.info(msg)