zabbix HA on AWS

2017-03-16 KasperDeng 更多博文 » 博客 » GitHub »

zabbix devops

原文链接 https://kasperdeng.github.io/zabbix-ha-on-aws/
注:以下为加速网络访问所做的原文缓存,经过重新格式化,可能存在格式方面的问题,或偶有遗漏信息,请以原文为准。


Zabbix HA Requirment

  • On AWS
  • 两台Zabbix Server, 在AWS上,处于一个VPC,但分别处在两个不同的AZ。
  • Active-Passive mode. 原因:zabbix server 自身有一套external scrips 去AWS cloudwatch 获取AWS service metrics。 如果是采用Active-Active的方式的,两台Active Zabbix server会造成数据冗余或冲突。

HA Implementation

Decouple Zabbix Database

  • Zabbix存储的数据库需要解耦出来,方便Zabbix Server主从切换,保持数据存储的唯一性。
  • 对于Zabbix server来说,最重要的就是监控的数据。在AWS上,最方便的方式就是采用AWS的RDS作为Zabbix的后端数据库,利用AWS的服务。
  • AWS上RDS比较强大,有MySQL, MariaDB和Postgresql可供选择。为了HA方案在别的云平台上的可重用性,选择Postgresql最为后端数据库。

RDS Setup

  • 启用RDS前,设置好master-user-name和master-user-password
  • 选择RDS的Multi-AZ功能。备份,维护按需设置。
  • 估算好检测数据所需的数据库容量大小,选择对应的DB实例类型 (e.g. db.m4.large)
  • 更新RDS的security group的设置,保证与zabbix的连通性。

MariaDB

  • Verify connection
mysql -h <RDS-DNS> -P 3306 -u <RDS-master-user-name> -p
  • yum install -y mariadb which use its client to connect RDS
  • setsebool -P httpd_can_network_connect_db=1 and setsebool -P zabbix_can_network=1

Postgres

  • Verify connection
psql \
   --host=occ-<rds instance DNS name> \
   --port=5432 \
   --username <username> \
   --password
  • yum install -y postgresql
  • setsebool -P httpd_can_network_connect_db=1 and setsebool -P zabbix_can_network=1

AWS ELB

在AWS上,zabbix service是通过AWS ELB来发现后端哪个Zone上的Zabbix server处于服务状态。

  • 端口Health-Check:

    • external ELB:
    • 80 (apache)
    • 3000 (grafana)
    • internal ELB: 10051 (用于zabbix-agent上报指标)
  • NOTE: 在AWS上不能通过VIP来做,因为AWS上不支持传播多播包 multicast packet

Pacemaker

通过pacemaker(foss, scalable HA cluster resource manager)来提供HA的功能。

  • 具体安装,大体配置步骤,请参考 Configuring High Availability (HA) Zabbix Server on CentOS 7, 本文不累赘,但注意以下几点:

  • 对apache的状态检查,可以通过启用自带的状态模块(built-in module在/usr/lib64/httpd/modules/),配置如下:

    • apache httpd status configuration
    • Add LoadModule status_module modules/mod_status.so to /etc/httpd/conf/httpd.conf
    • Add url handling as below
# Allow server status reports generated by mod_status,
# with the URL of http://servername/server-status
# Change the ".example.com" to match your domain to enable.
#
<Location /server-status>
    SetHandler server-status
    Order deny,allow
    Deny from all
    Allow from all
</Location>
  • 在pacemaker里面创建资源
pcs resource create zabbix_server systemd:zabbix-server op monitor interval=10s
pcs resource create zabbix_httpd systemd:httpd configfile=/etc/httpd/conf/httpd.conf statusurl="http://127.0.0.1/server-status" op monitor interval=20s
pcs resource create grafana_server systemd:grafana-server op monitor interval=10s
  • 在pacemaker里面把3个resources(zabbix, apache, grafana)组成创建资源组
pcs resource group add zabbixcluster zabbix_server zabbix_httpd
  • 在AWS上注意网络连通性问题,尤其当用pcs status看到集群状态不对的时候。检查下security group里面以下端口是否开放了。
    • UDP 5405: corosync 用于集群状态检查

zabbix-ha-on-aws

MISC

  • config files:
    • /etc/corosync/corosync.conf
    • /var/lib/pcsd/pcs_settings.conf
  • logs
    • /var/log/pacemaker.log
    • /var/log/cluster/corosync.log
    • /var/log/pcsd/pcsd.log

Reference