HACMP的心跳簡(jiǎn)介

1、HACM啲心跳簡(jiǎn)介Heartbeating用于監(jiān)控網(wǎng)絡(luò)接口、通訊設(shè)備、IPlabel（servie,non-service,persistent）的可用性，以及節(jié)點(diǎn)的可用性。從HACMP5.開始，HA只使用基于RSC拓?fù)浞?wù)的心跳，經(jīng)典心跳不再使用（經(jīng)典心跳使用網(wǎng)絡(luò)接口模塊NIMs,直接由cIstrmgrES監(jiān)控）。HA通過在每個(gè)節(jié)點(diǎn)之間，在每個(gè)通訊接口和設(shè)備上交換消息來實(shí)現(xiàn)心跳。每個(gè)節(jié)點(diǎn)按指定的時(shí)間間隔向其他節(jié)點(diǎn)發(fā)送心跳信號(hào)，并期望在指定的時(shí)間間隔內(nèi)收到相應(yīng)節(jié)點(diǎn)的心跳信號(hào)。如果沒有收到心跳信號(hào)，則RSCT認(rèn)為發(fā)生錯(cuò)誤，報(bào)告給HACMP由HACM采取相應(yīng)的恢復(fù)措施。心跳信息可以通過2種途徑交換

2、：基于IP網(wǎng)絡(luò)基于non-IP網(wǎng)絡(luò)Cluster孤島：由于TCPIP網(wǎng)絡(luò)的原因（交換機(jī)、路由器、HUB，基于IP的心跳不能正常發(fā)送接收，如果沒有其他non-ip的心跳交換，每個(gè)節(jié)點(diǎn)都會(huì)認(rèn)為其他節(jié)點(diǎn)失敗，自己請(qǐng)求獲得資源，這將影響數(shù)據(jù)的一致性和完整性，所以HACMP應(yīng)該能夠區(qū)分是IP網(wǎng)絡(luò)故障還是節(jié)點(diǎn)故障。防止孤島的出現(xiàn)。NON-IP網(wǎng)絡(luò)心跳不使用TCPIP網(wǎng)絡(luò)傳輸心跳，所以能有效的避免由于TCPIP網(wǎng)絡(luò)故障造成的Cluster孤島?；诖疟P的心跳HACMP5.1以后才支持。此類型的心跳支持SSASCS、FC類型的存儲(chǔ)，使用磁盤（diskhb）交換心跳信息。該磁盤需要屬于增強(qiáng)的concurrent

3、vg。同時(shí)，此磁盤也可以用于存儲(chǔ)其他共享信息。1、一塊盤屬于一個(gè)網(wǎng)絡(luò)（2節(jié)點(diǎn)），2個(gè)節(jié)點(diǎn)上該磁盤的ID要一致2、每對(duì)節(jié)點(diǎn)配置一個(gè)網(wǎng)絡(luò)3、該磁盤需要是增強(qiáng)的concurrentvg的一部分，但和RG無關(guān)基于IP別名的心跳使用基于IP別名的心跳，當(dāng)HACM啟動(dòng)時(shí)，在每個(gè)存在的IP上添加一個(gè)IP別名用于心跳信息交換，該別名需要使用不同的子網(wǎng)，并且不屬于任何名字解析。RSCT使用該別名為每個(gè)通訊接口建立通訊組（心跳環(huán)），來交換心跳信息。該方式的心跳不再監(jiān)控baseIP地址，而監(jiān)視通訊接口和serviceIP。IP別名的子網(wǎng)掩碼需要和sercieIP的子網(wǎng)掩碼相同。配置基于IP別名心跳的HACM，你需要

4、指定起始的IP地址。例如：一個(gè)2節(jié)點(diǎn)的HA每個(gè)節(jié)點(diǎn)2個(gè)網(wǎng)絡(luò)接口enO、en1,起始用于心跳的IP別名是則使用IP別名的適配器存儲(chǔ)在HACMPadapterOD類中Adapter/NodeNode1Node2ringen0Ring1en1Ring2心跳通訊測(cè)試系統(tǒng)環(huán)境：H80(OS52000於HA5.3)F50(OS520008PHA5.3)FAStT600心跳配置：網(wǎng)絡(luò)別名心跳，心跳別名初始化IP:串口心跳，分別連主機(jī)的串口3-dtty1磁盤心跳，增強(qiáng)的并行vg,hbvg-dhdisk4,FAStT存儲(chǔ)測(cè)試：1、在2個(gè)節(jié)點(diǎn)啟動(dòng)HA2、在主節(jié)點(diǎn)執(zhí)行l(wèi)ssrc-lstopsvcs#lssrc-ls

5、topsvcsSubsystemGrouptustopsvcstopsvcsNetworkNameIndxDefdMbrsStAdapterIDSubsystemGrouptustopsvcstopsvcsNetworkNameIndxDefdMbrsStAdapterIDnet_ether_01_00PIDSta491574activeGroupID22S0xnet_ether_01_00en10x4508653bHBInterval=1.000secs.Sensitivity=10missedbeatsMissedHBs:Total:0Currentgroup:0Packetssent:2

6、49ICMP0Errors:0Nombuf:0Packetsreceived:387ICMP0Dropped:0NIM'sPID:454686net_ether_01_1122Snet_ether_01_11en00x4508653c0xHBInterval=1.000secs.Sensitivity=10missedbeatsMissedHBs:Total:0Currentgroup:0Packetssent:249ICMP0Errors:0Nombuf:0Packetsreceived:387ICMP0Dropped:0NIM'sPID:503972rs232_0222SH

7、BInterval=2.000secs.Sensitivity=5missedbeatsrs232_02tty10x8508656b0x8508656eMissedHBs:Total:0Currentgroup:0Packetssent:186ICMP0Errors:0Nombuf:0Packetsreceived:178ICMP0Dropped:0NIM'sPID:532510diskhb_O322Sdiskhb_O3rhdisk40x8508653a0x8508656cHBInterval=2.000secs.Sensitivity=4missedbeatsMissedHBs:To

8、tal:0Currentgroup:0Packetssent:126ICMP0Errors:0Nombuf:0Packetsreceived:125ICMP0Dropped:0NIM'sPID:4342282locallyconnectedClientswithPIDs:haemd(413824)hagsd(450668)DeadManSwitchEnabled:resetinterval=1secondstripinterval=20secondsConfigurationInstance=3DaemonemploysnosecuritySegmentspinned:TextData

9、.Textsegmentsize:767KB.Staticdatasegmentsize:957KB.Dynamicdatasegmentsize:4233.Numberofoutstandingmalloc:222Usertime0sec.Systemtime0sec.Numberofpagefaults:263.Processswappedout0times.Numberofnodesup:2.Numberofnodesdown:0.用于心跳的進(jìn)程#ps-ef|grepnimroot434228491574015:08:23-0:00/usr/sbin/rsct/bin/hats_disk

10、hb_nimroot454686491574015:08:23-0:00/usr/sbin/rsct/bin/hats_nimroot503972491574015:08:23-0:00/usr/sbin/rsct/bin/hats_nimroot532510491574115:08:23-0:01/usr/sbin/rsct/bin/hats_rs232_nim3、備份節(jié)點(diǎn)上運(yùn)行l(wèi)ssrc-lstopsvcsStaactivGroupID#lssrc-lstopsvcsSubsystemGroupPIDtustopsvcstopsvcs29482eNetworkNameIndxDefdMbr

11、sStAdapterIDnet_ether_01_0022Sen1en1OxOxHBInterval=1.000secs.Sensitivity=10missedbeatsMissedHBs:Total:1Currentgroup:1Packetssent:1174ICMP0Errors:0Nombuf:0Packetsreceived:1726ICMP0Dropped:0NIM'sPID:28006net_ether_01_1122Snet_ether_01_11en00x0xHBInterval=1.000secs.Sensitivity=10missedbeatsMissedHB

12、s:Total:0Currentgroup:0Packetssent:1175ICMP0Errors:0Nombuf:0Packetsreceived:1724ICMP0Dropped:0NIM'sPID:27640rs232_0222Srs232_02tty10x0x8508656eHBInterval=2.000secs.Sensitivity=5missedbeatsMissedHBs:Total:1Currentgroup:1Packetssent:13173ICMP0Errors:0Nombuf:0Packetsreceived:810ICMP0Dropped:0NIM

13、9;sPID:25294diskhb_0322Sdiskhb_03rhdisk40x0x8508656cHBInterval=2.000secs.Sensitivity=4missedbeatsMissedHBs:Total:0Currentgroup:0Packetssent:568ICMP0Errors:0Nombuf:0Packetsreceived:570ICMP0Dropped:0NIM'sPID:268922locallyconnectedClientswithPIDs:haemd(26836)hagsd(27390)DeadManSwitchEnabled:resetin

14、terval=1secondstripinterval=20secondsConfigurationInstance=3DaemonemploysnosecuritySegmentspinned:TextData.Textsegmentsize:767KB.Staticdatasegmentsize:957KB.Dynamicdatasegmentsize:4169.Numberofoutstandingmalloc:222Usertime1sec.Systemtime3sec.Numberofpagefaults:423.Processswappedout0times.Numberofnod

15、esup:2.Numberofnodesdown:0.根據(jù)上面的信息可以看出，一共有4個(gè)心跳環(huán)在傳輸心跳信號(hào)，2個(gè)以太網(wǎng)，1個(gè)串口，1個(gè)磁盤。心跳信號(hào)在心跳環(huán)內(nèi)進(jìn)行傳輸。我們也可以通過日志來查看心跳傳輸?shù)那闆r：/var/ha/log目錄下的，3、down網(wǎng)卡enOHA進(jìn)行了正常的網(wǎng)卡swap操作。此時(shí)查看心跳日志。09/1410:22:05.356:ErrorsendingtoBadfilenumber.09/1410:22:05.356:Dispatchingnetmonrequestwhileanotherinprogress.09/1410:22:05.356:ReceivedaSEND

16、MSGcommand.Dst:09/1410:22:05.37609/1410:22:05.37609/1410:22:05.37609/1410:22:05.37609/1410:22:05.37609/1410:22:08.53809/1410:22:08.538ErrorsendingtoErrorsendingtoErrorsendingtoErrorsendingtoErrorsendingtoNetworkisdown.Networkisdown.Networkisdown.Networkisdown.Nmonresponse:AdapterisdownAdapterstatussuccessfullysent.此時(shí)en0不再發(fā)送心跳信息。備份節(jié)點(diǎn)的en0發(fā)現(xiàn)發(fā)送給，并收到停止發(fā)送心跳信息的命令，隨后發(fā)送心跳信息的地址變成啟動(dòng)網(wǎng)卡en0后，心跳有開始正常傳輸。4、其他心跳環(huán)類似。5、更改心跳相關(guān)的參數(shù)ExtendedConfiguration-aExtendedTopologyConfiguration-aConfigureHACMPNetworkModules-aChangeaNetworkModuleusingPredefinedValues分別選擇ether,diskhb,rs232ChangeaClusterNetworkMod