heartbeat + pacemaker + postgres_streaming_replication
說明：
該文檔用於說明以hearbeat+pacemaker的方式實現PostgreSQL流複製自動切換。注意內容包括有關hearbeat/pacemaker知識總結以及整個環境的搭建過程和問題處理。
一、介紹
Heartbeat
自3版本開始，heartbeat將原來項目拆分爲了多個子項目（即多個獨立組件），現在的組件包括：heartbeat、cluster-glue、resource-agents。
各組件主要功能：
heartbeat：屬於集羣的信息層，負責維護集羣中所有節點的信息以及各節點之間的通信。
cluster-glue：包括LRM（本地資源管理器）、STONITH，將heartbeat與crm（集羣資源管理器）聯繫起來，屬於一箇中間層。
resource-agents：即各種資源腳本，由LRM調用從而實現各個資源的啓動、停止、監控等。
Heartbeat內部組件關係圖：
Pacemaker
Pacemaker，即Cluster Resource Manager（CRM），管理整個HA，客戶端通過pacemaker管理監控整個集羣。
常用的集羣管理工具：
（1）基於命令行
crm shell/pcs
（2）基於圖形化
pygui/hawk/lcmc/pcs
Hawk:http://clusterlabs.org/wiki/Hawk
Lcmc:http://www.drbd.org/mc/lcmc/
Pacemaker內部組件、模塊關係圖：
二、環境
2.1 OS
# cat /etc/issue CentOS release 6.4 (Final) Kernel \r on an \m # uname -a Linux node1 2.6.32-358.el6.x86_64 #1 SMP Fri Feb 22 00:31:26 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux
2.2 IP
node1:
eth0 192.168.100.161/24 GW 192.168.100.1 ---真實地址
eth1 2.2.2.1/24 ---心跳地址
eth2 192.168.2.1/24 ---流複製地址
node2:
eth0 192.168.100.162/24 GW 192.168.100.1 ---真實地址
eth1 2.2.2.2/24 ---心跳地址
eth2 192.168.2.2/24 ---流複製地址
虛擬地址：
eth0:0 192.168.100.163/24 ---vip-master
eth0:0 192.168.100.164/24 ---vip-slave
eth2:0 192.168.2.3/24 ---vip-rep
2.3 軟件版本
# rpm -qa | grep heartbeat heartbeat-devel-3.0.3-2.3.el5 heartbeat-debuginfo-3.0.3-2.3.el5 heartbeat-3.0.3-2.3.el5 heartbeat-libs-3.0.3-2.3.el5 heartbeat-devel-3.0.3-2.3.el5 heartbeat-3.0.3-2.3.el5 heartbeat-debuginfo-3.0.3-2.3.el5 heartbeat-libs-3.0.3-2.3.el5 # rpm -qa | grep pacemaker pacemaker-libs-1.0.12-1.el5.centos pacemaker-1.0.12-1.el5.centos pacemaker-debuginfo-1.0.12-1.el5.centos pacemaker-debuginfo-1.0.12-1.el5.centos pacemaker-1.0.12-1.el5.centos pacemaker-libs-1.0.12-1.el5.centos pacemaker-libs-devel-1.0.12-1.el5.centos pacemaker-libs-devel-1.0.12-1.el5.centos # rpm -qa | grep resource-agent resource-agents-1.0.4-1.1.el5 # rpm -qa | grep cluster-glue cluster-glue-libs-1.0.6-1.6.el5 cluster-glue-libs-1.0.6-1.6.el5 cluster-glue-1.0.6-1.6.el5 cluster-glue-libs-devel-1.0.6-1.6.el5
PostgreSQL Version：9.1.4
三、安裝
3.1 設置YUM源
# wget -O /etc/yum.repos.d/pacemaker.repo http://clusterlabs.org/rpm/epel-5/clusterlabs.repo
3.2 安裝heartbeat/pacemaker
安裝libesmtp：
# wget ftp://ftp.univie.ac.at/systems/linux/fedora/epel/5/x86_64/libesmtp-1.0.4-5.el5.x86_64.rpm # wget ftp://ftp.univie.ac.at/systems/linux/fedora/epel/5/i386/libesmtp-1.0.4-5.el5.i386.rpm # rpm -ivh libesmtp-1.0.4-5.el5.x86_64.rpm # rpm -ivh libesmtp-1.0.4-5.el5.i386.rpm
安裝pacemaker corosync:
# yum install heartbeat* pacemaker*
通過命令查看資源腳本：
# crm ra list ocf AoEtarget AudibleAlarm CTDB ClusterMon Delay Dummy EvmsSCC Evmsd Filesystem HealthCPU HealthSMART ICP IPaddr IPaddr2 IPsrcaddr IPv6addr LVM LinuxSCSI MailTo ManageRAID ManageVE Pure-FTPd Raid1 Route SAPDatabase SAPInstance SendArp ServeRAID SphinxSearchDaemon Squid Stateful SysInfo SystemHealth VIPArip VirtualDomain WAS WAS6 WinPopup Xen Xinetd anything apache conntrackd controld db2 drbd eDir88 exportfs fio iSCSILogicalUnit iSCSITarget ids iscsi jboss ldirectord mysql mysql-proxy nfsserver nginx o2cb oracle oralsnr pgsql ping pingd portblock postfix proftpd rsyncd scsi2reservation sfex syslog-ng tomcat vmware
禁止開機啓動：
# chkconfig heartbeat off
3.3 安裝PostgreSQL
安裝目錄爲/opt/pgsql
{安裝過程略}
爲postgres用戶配置環境變量：
[postgres@node1 ~]$ cat .bash_profile # .bash_profile # Get the aliases and functions if [ -f ~/.bashrc ]; then . ~/.bashrc fi # User specific environment and startup programs export PATH=/opt/pgsql/bin:$PATH:$HOME/bin export PGDATA=/opt/pgsql/data export PGUSER=postgres export PGPORT=5432 export LD_LIBRARY_PATH=/opt/pgsql/lib:$LD_LIBRARY_PATH
四、配置
4.1 hosts設置
# vim /etc/hosts 192.168.100.161 node1 192.168.100.162 node2
4.2 配置heartbeat
創建配置文件：
# cp /usr/share/doc/heartbeat-3.0.3/ha.cf /etc/ha.d/
修改配置：
# vim /etc/ha.d/ha.cf logfile /var/log/ha-log logfacility local0 keepalive 2 deadtime 30 warntime 10 initdead 120 ucast eth1 2.2.2.2 #node2上改爲2.2.2.1 auto_failback off node node1 node node2 pacemaker respawn #自heartbeat-3.0.4改爲crm respawn
4.3 生成密鑰
# (echo -ne "auth 1\n1 sha1 ";dd if=/dev/urandom bs=512 count=1 | openssl sha1 ) > /etc/ha.d/authkeys 1+0 records in 1+0 records out 512 bytes (512 B) copied, 0.00032444 s, 1.6 MB/s # chmod 0600 /etc/ha.d/authkeys
4.4 同步配置
[root@node1 ~]# cd /etc/ha.d/ [root@node1 ha.d]# scp authkeys ha.cf node2:/etc/ha.d/
4.5 下載替換腳本
pgsql腳本過舊，不支持配置pgsql.crm中設置的一些參數，需要從網上下載並替換pgsql
下載地址：
https://github.com/ClusterLabs/resource-agents
# unzip resource-agents-master.zip # cd resource-agents-master/heartbeat/ # cp pgsql /usr/lib/ocf/resource.d/heartbeat/ # cp ocf-shellfuncs.in /usr/lib/ocf/lib/heartbeat/ocf-shellfuncs # cp ocf-rarun /usr/lib/ocf/lib/heartbeat/ocf-rarun # chmod 755 /usr/lib/ocf/resource.d/heartbeat/pgsql
修改ocf-shellfuncs：
if [ -z "$OCF_ROOT" ]; then
# : ${OCF_ROOT=@OCF_ROOT_DIR@}
: ${OCF_ROOT=/usr/lib/ocf}
Fi
pgsql資源腳本特性：
●主節點失效切換
master宕掉時，RA檢測到該問題並將master標記爲stop，隨後將slave提升爲新的master。
●異步與同步切換
如果slave宕掉或者LAN中存在問題，那麼當設置爲同步複製時包含寫操作的事務將會被終止，也就意味着服務將停止。因此，爲防止服務停止RA將會動態地將同步轉換爲異步複製。
●初始啓動時自動識別新舊數據
當兩個或多個節點上的Pacemaker同時初始啓動時，RA通過每個節點上最近的replay location進行比較，找出最新數據節點。這個擁有最新數據的節點將被認爲是master。當然，若在一個節點上啓動pacemaker或者該節點上的pacemaker是第一個被啓動的，那麼它也將成爲master。RA依據停止前的數據狀態進行裁定。
●讀負載均衡
由於slave節點可以處理只讀事務，因此對於讀操作可以通過虛擬另一個虛擬IP來實現讀操作的負載均衡。
4.6 啓動heartbeat
啓動：
[root@node1 ~]# service heartbeat start [root@node2 ~]# service heartbeat start
檢測狀態：
[root@node1 ~]# crm status ============ Last updated: Fri Jan 24 08:02:54 2014 Stack: Heartbeat Current DC: node2 (43a4f083-c5d3-4c66-a387-b05d79b5dd89) - partition with quorum Version: 1.0.12-unknown 2 Nodes configured, unknown expected votes 0 Resources configured. ============ Online: [ node1 node2 ]
{heartbeat啓動成功}
測試：
禁用stonith，創建一個虛擬ip資源vip
[root@node1 ~]# crm configure property stonith-enabled="false" [root@node1 ~]# crm configure crm(live)configure# primitive vip ocf:heartbeat:IPaddr2 \ > params \ > ip="192.168.100.90" \ > nic="eth0" \ > cidr_netmask="24" \ > op start timeout="60s" interval="0s" on-fail="stop" \ > op monitor timeout="60s" interval="10s" on-fail="restart" \ > op stop timeout="60s" interval="0s" on-fail="block" crm(live)configure# commit crm(live)configure# quit bye
[root@node2 heartbeat]# crm_mon -1 ============ Last updated: Fri Jan 24 08:23:09 2014 Stack: Heartbeat Current DC: node2 (43a4f083-c5d3-4c66-a387-b05d79b5dd89) - partition with quorum Version: 1.0.12-unknown 2 Nodes configured, unknown expected votes 1 Resources configured. ============ Online: [ node1 node2 ] vip (ocf::heartbeat:IPaddr2): Started node1
{vip資源在node1上運行}
[root@node1 ~]# ping 192.168.100.90 PING 192.168.100.90 (192.168.100.90) 56(84) bytes of data. 64 bytes from 192.168.100.90: icmp_seq=1 ttl=64 time=0.060 ms 64 bytes from 192.168.100.90: icmp_seq=2 ttl=64 time=0.111 ms 64 bytes from 192.168.100.90: icmp_seq=3 ttl=64 time=0.123 ms
模擬node1故障
[root@node1 ~]# service heartbeat stop [root@node2 heartbeat]# crm_mon -1 ============ Last updated: Fri Jan 24 08:22:22 2014 Stack: Heartbeat Current DC: node2 (43a4f083-c5d3-4c66-a387-b05d79b5dd89) - partition with quorum Version: 1.0.12-unknown 2 Nodes configured, unknown expected votes 1 Resources configured. ============ Online: [ node2 ] OFFLINE: [ node1 ] vip (ocf::heartbeat:IPaddr2): Started node2
{node2順利接管資源vip}
重新恢復node1
[root@node1 ~]# service heartbeat start [root@node2 heartbeat]# crm_mon -1 ============ Last updated: Fri Jan 24 08:23:09 2014 Stack: Heartbeat Current DC: node2 (43a4f083-c5d3-4c66-a387-b05d79b5dd89) - partition with quorum Version: 1.0.12-unknown 2 Nodes configured, unknown expected votes 1 Resources configured. ============ Online: [ node1 node2 ] vip (ocf::heartbeat:IPaddr2): Started node1
{node1順利收回vip資源的接管權}
刪除資源：
[root@node1 ~]# crm_resource -D -r vip -t primitive [root@node1 ~]# crm_resource -L NO resources configured
4.7 配置流複製
在node1/node2上配置postgresql.conf/pg_hba.conf：
postgresql.conf : listen_addresses = '*' port = 5432 wal_level = hot_standby archive_mode = on archive_command = 'test ! -f /opt/archivelog/%f && cp %p /opt/archivelog/%f' max_wal_senders = 4 wal_keep_segments = 50 hot_standby = on pg_hba.conf : host replication postgres 192.168.2.0/24 trust
4.8 配置pacemaker
{關於pacemaker的配置可通過多種方式，如crmsh、hb_gui、pcs等，該實驗使用crmsh配置}
啓動node1的heartbeat，關閉node2的heartbeat。
編寫crm配置腳本：
[root@node1 ~]# cat pgsql.crm property \ no-quorum-policy="ignore" \ stonith-enabled="false" \ crmd-transition-delay="0s" rsc_defaults \ resource-stickiness="INFINITY" \ migration-threshold="1" ms msPostgresql pgsql \ meta \ master-max="1" \ master-node-max="1" \ clone-max="2" \ clone-node-max="1" \ notify="true" clone clnPingCheck pingCheck group master-group \ vip-master \ vip-rep \ meta \ ordered="false" primitive vip-master ocf:heartbeat:IPaddr2 \ params \ ip="192.168.100.163" \ nic="eth0" \ cidr_netmask="24" \ op start timeout="60s" interval="0s" on-fail="stop" \ op monitor timeout="60s" interval="10s" on-fail="restart" \ op stop timeout="60s" interval="0s" on-fail="block" primitive vip-rep ocf:heartbeat:IPaddr2 \ params \ ip="192.168.2.3" \ nic="eth2" \ cidr_netmask="24" \ meta \ migration-threshold="0" \ op start timeout="60s" interval="0s" on-fail="restart" \ op monitor timeout="60s" interval="10s" on-fail="restart" \ op stop timeout="60s" interval="0s" on-fail="block" primitive vip-slave ocf:heartbeat:IPaddr2 \ params \ ip="192.168.100.164" \ nic="eth0" \ cidr_netmask="24" \ meta \ resource-stickiness="1" \ op start timeout="60s" interval="0s" on-fail="restart" \ op monitor timeout="60s" interval="10s" on-fail="restart" \ op stop timeout="60s" interval="0s" on-fail="block" primitive pgsql ocf:heartbeat:pgsql \ params \ pgctl="/opt/pgsql/bin/pg_ctl" \ psql="/opt/pgsql/bin/psql" \ pgdata="/opt/pgsql/data/" \ start_opt="-p 5432" \ rep_mode="sync" \ node_list="node1 node2" \ restore_command="cp /opt/archivelog/%f %p" \ primary_conninfo_opt="keepalives_idle=60 keepalives_interval=5 keepalives_count=5" \ master_ip="192.168.2.3" \ stop_escalate="0" \ op start timeout="60s" interval="0s" on-fail="restart" \ op monitor timeout="60s" interval="7s" on-fail="restart" \ op monitor timeout="60s" interval="2s" on-fail="restart" role="Master" \ op promote timeout="60s" interval="0s" on-fail="restart" \ op demote timeout="60s" interval="0s" on-fail="stop" \ op stop timeout="60s" interval="0s" on-fail="block" \ op notify timeout="60s" interval="0s" primitive pingCheck ocf:pacemaker:pingd \ params \ name="default_ping_set" \ host_list="192.168.100.1" \ multiplier="100" \ op start timeout="60s" interval="0s" on-fail="restart" \ op monitor timeout="60s" interval="10s" on-fail="restart" \ op stop timeout="60s" interval="0s" on-fail="ignore" location rsc_location-1 vip-slave \ rule 200: pgsql-status eq "HS:sync" \ rule 100: pgsql-status eq "PRI" \ rule -inf: not_defined pgsql-status \ rule -inf: pgsql-status ne "HS:sync" and pgsql-status ne "PRI" location rsc_location-2 msPostgresql \ rule -inf: not_defined default_ping_set or default_ping_set lt 100 colocation rsc_colocation-1 inf: msPostgresql clnPingCheck colocation rsc_colocation-2 inf: master-group msPostgresql:Master order rsc_order-1 0: clnPingCheck msPostgresql order rsc_order-2 0: msPostgresql:promote master-group:start symmetrical=false order rsc_order-3 0: msPostgresql:demote master-group:stop symmetrical=false
導入配置腳本：
[root@node1 ~]# crm configure load update pgsql.crm WARNING: pgsql: specified timeout 60s for stop is smaller than the advised 120 WARNING: pgsql: specified timeout 60s for start is smaller than the advised 120 WARNING: pgsql: specified timeout 60s for notify is smaller than the advised 90 WARNING: pgsql: specified timeout 60s for demote is smaller than the advised 120 WARNING: pgsql: specified timeout 60s for promote is smaller than the advised 120 WARNING: pingCheck: specified timeout 60s for start is smaller than the advised 90 WARNING: pingCheck: specified timeout 60s for stop is smaller than the advised 100
一段時間後查看ha狀態：
[root@node1 ~]# crm_mon -Afr1 ============ Last updated: Mon Jan 27 05:23:59 2014 Stack: Heartbeat Current DC: node1 (30b7dc95-25c5-40d7-b1e4-7eaf2d5cdf07) - partition with quorum Version: 1.0.12-unknown 2 Nodes configured, unknown expected votes 4 Resources configured. ============ Online: [ node1 ] OFFLINE: [ node2 ] Full list of resources: vip-slave (ocf::heartbeat:IPaddr2): Started node1 Resource Group: master-group vip-master (ocf::heartbeat:IPaddr2): Started node1 vip-rep (ocf::heartbeat:IPaddr2): Started node1 Master/Slave Set: msPostgresql Masters: [ node1 ] Stopped: [ pgsql:0 ] Clone Set: clnPingCheck Started: [ node1 ] Stopped: [ pingCheck:1 ] Node Attributes: * Node node1: + default_ping_set : 100 + master-pgsql:1 : 1000 + pgsql-data-status : LATEST + pgsql-master-baseline : 0000000003000078 + pgsql-status : PRI Migration summary: * Node node1:
注：剛啓動時爲slave，一段時間後自動切換爲master。
待資源在node1上都正常運行後在node2上執行基礎備份同步：
[postgres@node2 data]$ pg_basebackup -h 192.168.2.3 -U postgres -D /opt/pgsql/data/ -P
啓動node2的heartbeat：
[root@node2 ~]# service heartbeat start
過一段時間後查看集羣狀態：
[root@node1 ~]# crm_mon -Afr1 ============ Last updated: Mon Jan 27 05:27:22 2014 Stack: Heartbeat Current DC: node1 (30b7dc95-25c5-40d7-b1e4-7eaf2d5cdf07) - partition with quorum Version: 1.0.12-unknown 2 Nodes configured, unknown expected votes 4 Resources configured. ============ Online: [ node1 node2 ] Full list of resources: vip-slave (ocf::heartbeat:IPaddr2): Started node2 Resource Group: master-group vip-master (ocf::heartbeat:IPaddr2): Started node1 vip-rep (ocf::heartbeat:IPaddr2): Started node1 Master/Slave Set: msPostgresql Masters: [ node1 ] Slaves: [ node2 ] Clone Set: clnPingCheck Started: [ node1 node2 ] Node Attributes: * Node node1: + default_ping_set : 100 + master-pgsql:1 : 1000 + pgsql-data-status : LATEST + pgsql-master-baseline : 0000000003000078 + pgsql-status : PRI * Node node2: + default_ping_set : 100 + master-pgsql:0 : 100 + pgsql-data-status : STREAMING|SYNC + pgsql-status : HS:sync Migration summary: * Node node2: * Node node1:
{vip-slave資源已經由node1切換到了node2上，流複製狀態正常}