千家信息网

Zabbix 监控 AlwaysOn

发表于:2025-01-20 作者:千家信息网编辑
千家信息网最后更新 2025年01月20日,基于生产环境 SQL Server 业务数据库的特性,需要针对 SQL Server 生产环境几组 AlwaysOn AG 的 Primary Replica 部署如下监控:1. Windows集群节
千家信息网最后更新 2025年01月20日Zabbix 监控 AlwaysOn

基于生产环境 SQL Server 业务数据库的特性,需要针对 SQL Server 生产环境几组 AlwaysOn AG 的 Primary Replica 部署如下监控:
1. Windows集群节点不正常
2. 任一业务数据库没加入可用性组
3. AG中任一数据库状态异常
4. 任一数据库延时超过阈值


1. primary replica 检查Windows集群状态不正常的节点数(>0 告警)

SELECT COUNT (*) as NotHealtyCountFROM sys.dm_hadr_cluster_membersWHERE member_state_desc !='UP';




2. primary replica 检查副本可用数据库个数 与 需要加入AG的业务数据库数 对比 (<后者告警)

SELECT COUNT(*) as DBCount FROM sys.availability_databases_cluster;




3. primary replica 检查任一数据库状态异常(>0 告警)

SELECT COUNT (*) as NotHealtyCountFROM [master].[sys].[dm_hadr_database_replica_states]WHERE [synchronization_health_desc] != N'HEALTHY'




4. primary replica 检查任一数据库延时(>900 告警)

;WITH AG_Stats AS (SELECT AR.replica_server_name,   HARS.role_desc,    Db_name(DRS.database_id) [DBName],    DRS.last_commit_timeFROM   sys.dm_hadr_database_replica_states DRS INNER JOIN sys.availability_replicas AR ON DRS.replica_id = AR.replica_id INNER JOIN sys.dm_hadr_availability_replica_states HARS ON AR.group_id = HARS.group_id AND AR.replica_id = HARS.replica_id ),Pri_CommitTime AS (SELECTreplica_server_name, DBName, last_commit_timeFROMAG_StatsWHERErole_desc = 'PRIMARY'),Sec_CommitTime AS (SELECTreplica_server_name, DBName, last_commit_timeFROMAG_StatsWHERErole_desc = 'SECONDARY')SELECT ISNULL(DATEDIFF(ss,s.last_commit_time,p.last_commit_time),0) AS [Sync_Lag_Secs]FROM Pri_CommitTime pLEFT JOIN Sec_CommitTime s ON [s].[DBName] = [p].[DBName]




Zabbix参数配置示例:

UserParameter=MSSQL.SyncHealth,"C:\Program Files\Microsoft SQL Server\Client SDK\ODBC\110\Tools\Binn\SQLCMD.EXE" -d Master -U zabbix -P zabbix -h -1 -W -Q "SELECT COUNT (*) as NotHealtyCount

FROM [master].[sys].[dm_hadr_database_replica_states]

WHERE [synchronization_health_desc] != N''HEALTHY''

"


建议,将1、3、4项配置为模板,2项需要单独配置4个参数。


0