monitor_ob.yaml

obInputBasic: &obInputBasic
  plugin: mysqlTableInput
  config:
    timeout: 10s
    pluginConfig:
      collect_interval: ${monagent.second.metric.cache.update.interval}
      connection:
        url: ${monagent.ob.monitor.user}:${monagent.ob.monitor.password}@tcp(127.0.0.1:${monagent.ob.sql.port})/oceanbase?interpolateParams=true
        maxIdle: 2
        maxOpen: 32
      defaultConditionValues:
        ob_svr_ip: ${monagent.host.ip}
        ob_svr_port: ${monagent.ob.rpc.port}
        ob_is_rootservice: true
      collectConfig:
        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ with_rootserver from __all_server where svr_ip = ? and svr_port = ?
          name: ob_role
          sqlSlowThreshold: 100ms
          params: [ ob_svr_ip, ob_svr_port ]
          minObVersion: ~
          maxObVersion: 4.0.0.0
          conditionValues:
            ob_is_rootservice: with_rootserver
          enableCache: true
          cacheExpire: 1m
        - sql: select (case when with_rootserver='YES' then 1 else 0 end) as with_rootserver from DBA_OB_SERVERS where svr_ip = ? and svr_port = ?
          name: ob_role
          sqlSlowThreshold: 100ms
          params: [ ob_svr_ip, ob_svr_port ]
          minObVersion: 4.0.0.0
          maxObVersion: ~
          conditionValues:
            ob_is_rootservice: with_rootserver
          enableCache: true
          cacheExpire: 1m

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ tenant_id, cache_name, cache_size from __all_virtual_kvcache_info where svr_ip = ? and svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_cache
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
            tenant_name: tenant_name
            cache_name: cache_name
          metrics:
            size_bytes: cache_size
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}
        - sql: select /* MONITOR_AGENT */ tenant_id, cache_name, cache_size from GV$OB_KVCACHE where svr_ip = ? and svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_cache
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            ob_tenant_id: tenant_id
            tenant_name: tenant_name
            cache_name: cache_name
          metrics:
            size_bytes: cache_size
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ con_id tenant_id, stat_id, value from v$sysstat where stat_id IN (10000, 10001, 10002, 10003, 10004, 10005, 10006, 140002, 140003, 140005, 140006, 40030, 80040, 80041, 130000, 130001, 130002, 130004, 20000, 20001, 20002, 30000, 30001, 30002, 30005, 30006, 30007, 30008, 30009, 30010, 30011, 30012, 30013, 30080, 30081, 40000, 40001, 40002, 40003, 40004, 40005, 40006, 40007, 40008, 40009, 40010, 40011, 40012, 40018, 40019, 40116, 40117, 40118, 50000, 50001, 50002, 50004, 50005, 50008, 50009, 50010, 50011, 50037, 50038, 60000, 60001, 60002, 60003, 60004, 60005, 60019, 60020, 60021, 60022, 60023, 60024, 80057, 120000, 120001, 120009, 120008) and (con_id > 1000 or con_id = 1) and class < 1000
          name: ob_sysstat
          sqlSlowThreshold: 100ms
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
            stat_id: stat_id
          metrics:
            value: value
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}
        - sql: select /* MONITOR_AGENT */ con_id tenant_id, stat_id, value from v$sysstat where stat_id IN (10000, 10001, 10002, 10003, 10004, 10005, 10006, 140002, 140003, 140005, 140006, 40030, 80040, 80041, 130000, 130001, 130002, 130004, 20000, 20001, 20002, 30000, 30001, 30002, 30005, 30006, 30007, 30008, 30009, 30010, 30011, 30012, 30013, 40000, 40001, 40002, 40003, 40004, 40005, 40006, 40007, 40008, 40009, 40010, 40011, 40012, 40018, 40019, 50000, 50001, 60087, 50004, 50005, 50008, 50009, 50010, 50011, 50037, 50038, 60000, 60001, 60002, 60003, 60004, 60005, 60019, 60020, 60021, 60022, 60023, 60024, 80057, 120000, 120001, 120009, 120008) and (con_id > 1000 or con_id = 1) and class < 1000
          name: ob_sysstat
          sqlSlowThreshold: 100ms
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            ob_tenant_id: tenant_id
            stat_id: stat_id
          metrics:
            value: value
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ group_concat(svr_ip SEPARATOR ',') as servers, status, count(1) as cnt from __all_server group by status
          name: ob_server
          sqlSlowThreshold: 100ms
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            server_ips: servers
            status: status
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: 60s
        - sql: select /* MONITOR_AGENT */ group_concat(svr_ip SEPARATOR ',') as servers, status, count(1) as cnt from DBA_OB_SERVERS group by status
          name: ob_server
          sqlSlowThreshold: 100ms
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            server_ips: servers
            status: status
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: 60s

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) QUERY_TIMEOUT(100000000) */ tenant_name, tenant_id, count(*)  as cnt from gv$table group by tenant_id
          name: ob_table
          sqlSlowThreshold: 100ms
          condition: ob_is_rootservice
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: 1h
        - sql: select /*+ MONITOR_AGENT QUERY_TIMEOUT(100000000) */ con_id tenant_id, count(*) as cnt from CDB_TABLES group by con_id
          name: ob_table
          sqlSlowThreshold: 100ms
          condition: ob_is_rootservice
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            ob_tenant_id: tenant_id
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: 1h

        - sql: select  /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ case when cnt is null then 0 else cnt end as cnt, tenant_name, tenant_id from (select __all_tenant.tenant_name, __all_tenant.tenant_id, cnt from __all_tenant left join (select count(`state`='ACTIVE' OR NULL) as cnt, tenant as tenant_name from __all_virtual_processlist where svr_ip = ? and svr_port = ? group by tenant) t1 on __all_tenant.tenant_name = t1.tenant_name) t2
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_active_session
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
            tenant_name: tenant_name
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}
        - sql: select /* MONITOR_AGENT */ case when cnt is null then 0 else cnt end as cnt, tenant_name, tenant_id from (select DBA_OB_TENANTS.tenant_name, DBA_OB_TENANTS.tenant_id, cnt from DBA_OB_TENANTS left join (select count(`state`='ACTIVE' OR NULL) as cnt, tenant as tenant_name from GV$OB_PROCESSLIST where svr_ip = ? and svr_port = ? group by tenant) t1 on DBA_OB_TENANTS.tenant_name = t1.tenant_name where DBA_OB_TENANTS.tenant_type<>'META') t2
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_active_session
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            ob_tenant_id: tenant_id
            tenant_name: tenant_name
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}

        - sql: select  /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ case when cnt is null then 0 else cnt end as cnt, tenant_name, tenant_id from (select __all_tenant.tenant_name, __all_tenant.tenant_id, cnt from __all_tenant left join (select count(1) as cnt, tenant as tenant_name from __all_virtual_processlist where svr_ip = ? and svr_port = ? group by tenant) t1 on __all_tenant.tenant_name = t1.tenant_name) t2
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_all_session
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
            tenant_name: tenant_name
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}
        - sql: select /* MONITOR_AGENT */ case when cnt is null then 0 else cnt end as cnt, tenant_name, tenant_id from (select DBA_OB_TENANTS.tenant_name, DBA_OB_TENANTS.tenant_id, cnt from DBA_OB_TENANTS left join (select count(1) as cnt, tenant as tenant_name from GV$OB_PROCESSLIST where svr_ip = ? and svr_port = ? group by tenant) t1 on DBA_OB_TENANTS.tenant_name = t1.tenant_name where DBA_OB_TENANTS.tenant_type<>'META') t2
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_all_session
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            ob_tenant_id: tenant_id
            tenant_name: tenant_name
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ tenant_id, mem_used, access_count, hit_count from v$plan_cache_stat
          name: ob_plan_cache
          sqlSlowThreshold: 100ms
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
          metrics:
            memory_bytes: mem_used
            access_total: access_count
            hit_total: hit_count
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}
        - sql: select /* MONITOR_AGENT */ tenant_id, mem_used, access_count, hit_count from V$OB_PLAN_CACHE_STAT
          name: ob_plan_cache
          sqlSlowThreshold: 100ms
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            ob_tenant_id: tenant_id
          metrics:
            memory_bytes: mem_used
            access_total: access_count
            hit_total: hit_count
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ con_id tenant_id, sum(total_waits) as total_waits, sum(time_waited_micro) / 1000000 as time_waited from v$system_event where v$system_event.wait_class <> 'IDLE' and (con_id > 1000 or con_id = 1) group by tenant_id
          name: ob_waitevent
          sqlSlowThreshold: 100ms
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
          metrics:
            wait_total: total_waits
            wait_seconds_total: time_waited
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}
        - sql: select /* MONITOR_AGENT */ con_id tenant_id, sum(total_waits) as total_waits, sum(time_waited_micro) / 1000000 as time_waited from v$system_event where v$system_event.wait_class <> 'IDLE' and (con_id > 1000 or con_id = 1) group by tenant_id
          name: ob_waitevent
          sqlSlowThreshold: 100ms
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            ob_tenant_id: tenant_id
          metrics:
            wait_total: total_waits
            wait_seconds_total: time_waited
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ con_id tenant_id, case when event_id = 10000 then 'INTERNAL' when event_id = 13000 then 'SYNC_RPC' when event_id = 14003 then 'ROW_LOCK_WAIT' when (event_id >= 10001 and event_id <= 11006) or (event_id >= 11008 and event_id <= 11011) then 'IO' when event like 'latch:%' then 'LATCH' else 'OTHER' END event_group, sum(total_waits) as total_waits, sum(time_waited_micro / 1000000) as time_waited from v$system_event where v$system_event.wait_class <> 'IDLE' and (con_id > 1000 or con_id = 1) group by tenant_id, event_group
          name: ob_system_event
          sqlSlowThreshold: 100ms
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
            event_group: event_group
          metrics:
            total_waits: total_waits
            time_waited: time_waited
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}
        - sql: select /* MONITOR_AGENT */ con_id tenant_id, case when event_id = 10000 then 'INTERNAL' when event_id = 13000 then 'SYNC_RPC' when event_id = 14003 then 'ROW_LOCK_WAIT' when (event_id >= 10001 and event_id <= 11006) or (event_id >= 11008 and event_id <= 11011) then 'IO' when event like 'latch:%' then 'LATCH' else 'OTHER' END event_group, sum(total_waits) as total_waits, sum(time_waited_micro / 1000000) as time_waited from v$system_event where v$system_event.wait_class <> 'IDLE' and (con_id > 1000 or con_id = 1) group by tenant_id, event_group
          name: ob_system_event
          sqlSlowThreshold: 100ms
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            ob_tenant_id: tenant_id
            event_group: event_group
          metrics:
            total_waits: total_waits
            time_waited: time_waited
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ cpu_total, cpu_assigned, mem_total, mem_assigned,disk_total, cpu_assigned_percent, mem_assigned_percent from __all_virtual_server_stat where svr_ip = ? and svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_server_resource
          minObVersion: ~
          maxObVersion: 2.0.0
          metrics:
            cpu: cpu_total
            cpu_assigned: cpu_assigned
            memory_bytes: mem_total
            memory_assigned_bytes: mem_assigned
            disk_bytes: disk_total
            cpu_assigned_percent: cpu_assigned_percent
            memory_assigned_percent: mem_assigned_percent
          enableCache: true
          cacheExpire: 60s
        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ cpu_total,cpu_max_assigned as cpu_assigned,mem_total,mem_max_assigned as mem_assigned,disk_total, cpu_assigned_percent, mem_assigned_percent from __all_virtual_server_stat where svr_ip = ? and svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_server_resource
          minObVersion: 2.0.0
          maxObVersion: 4.0.0.0
          metrics:
            cpu: cpu_total
            cpu_assigned: cpu_assigned
            memory_bytes: mem_total
            memory_assigned_bytes: mem_assigned
            disk_bytes: disk_total
            cpu_assigned_percent: cpu_assigned_percent
            memory_assigned_percent: mem_assigned_percent
          enableCache: true
          cacheExpire: 60s
        - sql: select /* MONITOR_AGENT */ cpu_capacity_max as cpu_total,cpu_assigned_max as cpu_assigned,mem_capacity as mem_total,mem_assigned as mem_assigned,data_disk_capacity as disk_total, (cpu_assigned_max / cpu_capacity_max) as cpu_assigned_percent, (mem_assigned / mem_capacity) as mem_assigned_percent from GV$OB_SERVERS where svr_ip = ? and svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_server_resource
          minObVersion: 4.0.0.0
          maxObVersion: ~
          metrics:
            cpu: cpu_total
            cpu_assigned: cpu_assigned
            memory_bytes: mem_total
            memory_assigned_bytes: mem_assigned
            disk_bytes: disk_total
            cpu_assigned_percent: cpu_assigned_percent
            memory_assigned_percent: mem_assigned_percent
          enableCache: true
          cacheExpire: 60s

        - sql: SELECT /*+read_consistency(weak) */ COALESCE(tenant_id, -1) as tenant_id, tenant_name, SUM(max_cpu) AS max_cpu, SUM(min_cpu) AS min_cpu, SUM(max_memory) AS max_memory, SUM(min_memory) AS min_memory FROM v$unit GROUP BY tenant_id
          sqlSlowThreshold: 100ms
          name: ob_tenant_resource
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            tenant_name: tenant_name
            ob_tenant_id: tenant_id
          metrics:
            max_cpu: max_cpu
            min_cpu: min_cpu
            max_memory: max_memory
            min_memory: min_memory
          enableCache: true
          cacheExpire: 60s
        - sql: select coalesce(t1.tenant_id, -1) as tenant_id, tenant_name, sum(max_cpu) as max_cpu, sum(min_cpu) as min_cpu, sum(max_memory) as max_memory, sum(min_memory) as min_memory from (select t1.unit_id, t1.svr_ip, t1.svr_port, t2.tenant_id, t1.min_cpu, t1.max_cpu, t1.min_memory, t1.max_memory from (select  unit_id, svr_ip, svr_port, sum(min_cpu) as min_cpu, sum(max_cpu) as max_cpu, sum(memory_size) as min_memory, sum(memory_size) as max_memory from v$ob_units  group by unit_id ) t1 join dba_ob_units t2 on t1.unit_id = t2.unit_id) t1 join dba_ob_tenants t2 on t1.tenant_id = t2.tenant_id where tenant_type <>'meta' group by tenant_id
          sqlSlowThreshold: 100ms
          name: ob_tenant_resource
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            tenant_name: tenant_name
            ob_tenant_id: tenant_id
          metrics:
            max_cpu: max_cpu
            min_cpu: min_cpu
            max_memory: max_memory
            min_memory: min_memory
          enableCache: true
          cacheExpire: 60s

        - sql: SELECT t2.tenant_id, t2.tenant_name, t1.cpu_total, t1.cpu_assigned, t1.mem_total, t1.mem_assigned FROM __all_virtual_server_stat t1 JOIN (SELECT tenant_id, tenant_name, svr_ip, svr_port FROM `gv$unit`) t2 ON t1.svr_ip=t2.svr_ip AND t1.svr_port=t2.svr_port WHERE t1.svr_ip = ? AND t1.svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_tenant_assigned
          minObVersion: ~
          maxObVersion: 2.0.0
          metrics:
            cpu_total: cpu_total
            cpu_assigned: cpu_assigned
            mem_total: mem_total
            mem_assigned: mem_assigned
          tags:
            tenant_name: tenant_name
            ob_tenant_id: tenant_id
          enableCache: true
          cacheExpire: 60s
        - sql: select /* MONITOR_AGENT */ t4.tenant_id, t4.tenant_name, cpu_capacity as cpu_total,cpu_assigned,mem_capacity as mem_total,mem_assigned as mem_assigned from GV$OB_SERVERS t1 JOIN (SELECT t2.tenant_id, t3.tenant_name, t2.svr_ip, t2.svr_port FROM gv$ob_units t2 left join DBA_OB_TENANTS t3 on t2.tenant_id=t3.tenant_id WHERE T3.tenant_type<>'META') t4 ON t1.svr_ip=t4.svr_ip AND t1.svr_port=t4.svr_port WHERE t1.svr_ip = ? AND t1.svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_tenant_assigned
          minObVersion: 4.0.0.0
          maxObVersion: ~
          metrics:
            cpu_total: cpu_total
            cpu_assigned: cpu_assigned
            mem_total: mem_total
            mem_assigned: mem_assigned
          tags:
            tenant_name: tenant_name
            ob_tenant_id: tenant_id
          enableCache: true
          cacheExpire: 60s

        - sql: SELECT t2.tenant_id, t2.tenant_name, t1.total_size FROM __all_virtual_disk_stat t1 JOIN (SELECT tenant_id, tenant_name, svr_ip, svr_port FROM `gv$unit`) t2 ON t1.svr_ip=t2.svr_ip AND t1.svr_port=t2.svr_port WHERE t1.svr_ip = ? AND t1.svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_tenant_disk
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            tenant_name: tenant_name
            ob_tenant_id: tenant_id
          metrics:
            total_size: total_size
          enableCache: true
          cacheExpire: 60s
        - sql: select coalesce(t1.tenant_id, -1) as tenant_id, tenant_name, sum(data_disk_in_use) as data_disk_in_use, sum(log_disk_in_use) as log_disk_in_use from (select t1.unit_id, t1.svr_ip, t1.svr_port, t2.tenant_id, t1.data_disk_in_use, t1.log_disk_in_use from (select  unit_id, svr_ip, svr_port, sum(data_disk_in_use) as data_disk_in_use, sum(log_disk_in_use) as log_disk_in_use from v$ob_units  group by unit_id ) t1 join dba_ob_units t2 on t1.unit_id = t2.unit_id) t1 join dba_ob_tenants t2 on t1.tenant_id = t2.tenant_id where tenant_type <>'meta' group by tenant_id
          sqlSlowThreshold: 100ms
          name: ob_tenant_disk
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            tenant_name: tenant_name
            ob_tenant_id: tenant_id
          metrics:
            data_size: data_disk_in_use
            log_size: log_disk_in_use
          enableCache: true
          cacheExpire: 60s

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ total_size, free_size from __all_virtual_disk_stat where svr_ip = ? and svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_disk
          minObVersion: ~
          maxObVersion: 4.0.0.0
          metrics:
            total_bytes: total_size
            free_bytes: free_size
          enableCache: true
          cacheExpire: 60s
        - sql: select /* MONITOR_AGENT */ data_disk_capacity as total_size, (data_disk_capacity - data_disk_in_use) as free_size from GV$OB_SERVERS where svr_ip = ? and svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_disk
          minObVersion: 4.0.0.0
          maxObVersion: ~
          metrics:
            total_bytes: total_size
            free_bytes: free_size
          enableCache: true
          cacheExpire: ${monagent.second.metric.cache.update.interval}

obInputExtra: &obInputExtra
  plugin: mysqlTableInput
  config:
    timeout: 10s
    pluginConfig:
      collect_interval: ${monagent.collector.ob.extra.interval}
      connection:
        url: ${monagent.ob.monitor.user}:${monagent.ob.monitor.password}@tcp(127.0.0.1:${monagent.ob.sql.port})/oceanbase?interpolateParams=true
        maxIdle: 2
        maxOpen: 32
      defaultConditionValues:
        ob_svr_ip: ${monagent.host.ip}
        ob_svr_port: ${monagent.ob.rpc.port}
        ob_is_rootservice: true
      collectConfig:
        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ with_rootserver from __all_server where svr_ip = ? and svr_port = ?
          name: ob_role
          sqlSlowThreshold: 100ms
          params: [ ob_svr_ip, ob_svr_port ]
          minObVersion: ~
          maxObVersion: 4.0.0.0
          conditionValues:
            ob_is_rootservice: with_rootserver
          enableCache: true
          cacheExpire: 1m
        - sql: select (case when with_rootserver='YES' then 1 else 0 end) as with_rootserver from DBA_OB_SERVERS where svr_ip = ? and svr_port = ?
          name: ob_role
          sqlSlowThreshold: 100ms
          params: [ ob_svr_ip, ob_svr_port ]
          minObVersion: 4.0.0.0
          maxObVersion: ~
          conditionValues:
            ob_is_rootservice: with_rootserver
          enableCache: true
          cacheExpire: 1m

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ count(*) as cnt from v$unit
          name: ob_unit
          sqlSlowThreshold: 100ms
          minObVersion: ~
          maxObVersion: 4.0.0.0
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: 60s
        - sql: select /* MONITOR_AGENT */ count(*) as cnt from V$OB_UNITS
          name: ob_unit
          sqlSlowThreshold: 100ms
          minObVersion: 4.0.0.0
          maxObVersion: ~
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: 60s

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) QUERY_TIMEOUT(100000000) */ count(*) as cnt from gv$table where table_type in (5) and index_status in (5, 6)
          name: ob_index
          sqlSlowThreshold: 100ms
          condition: ob_is_rootservice
          minObVersion: ~
          maxObVersion: 4.0.0.0
          metrics:
            error_num: cnt
          enableCache: true
          cacheExpire: 60s
        - sql: select /*+ MONITOR_AGENT QUERY_TIMEOUT(100000000) */ count(*) as cnt from CDB_INDEXES where status in ('ERROR','UNUSABLE')
          name: ob_index
          sqlSlowThreshold: 100ms
          condition: ob_is_rootservice
          minObVersion: 4.0.0.0
          maxObVersion: ~
          metrics:
            error_num: cnt
          enableCache: true
          cacheExpire: 60s

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ tenant_id,  active, total, freeze_trigger, freeze_cnt from gv$memstore where ip = ? and port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_memstore
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
          metrics:
            active_bytes: active
            total_bytes: total
            freeze_trigger_bytes: freeze_trigger
            freeze_times: freeze_cnt
          enableCache: true
          cacheExpire: 60s
        - sql: select /* MONITOR_AGENT */ tenant_id, active_span as active, memstore_used as total, freeze_trigger, freeze_cnt from GV$OB_MEMSTORE  where svr_ip = ? and svr_port = ?
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_memstore
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            ob_tenant_id: tenant_id
          metrics:
            active_bytes: active
            total_bytes: total
            freeze_trigger_bytes: freeze_trigger
            freeze_times: freeze_cnt
          enableCache: true
          cacheExpire: 60s

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ sum(hold) as hold, sum(used) as used from __all_virtual_memory_info where tenant_id = 500 and svr_ip = ? and svr_port = ? and mod_name <> 'OB_KVSTORE_CACHE_MB'
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_tenant500_memory
          minObVersion: ~
          maxObVersion: 4.0.0.0
          metrics:
            hold_bytes: hold
            used_bytes: used
          enableCache: true
          cacheExpire: 60s
        - sql: select /* MONITOR_AGENT */ sum(hold) as hold, sum(used) as used from GV$OB_MEMORY where tenant_id = 500 and svr_ip = ? and svr_port = ? and MOD_NAME <> 'KvstorCacheMb'
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_tenant500_memory
          minObVersion: 4.0.0.0
          maxObVersion: ~
          metrics:
            hold_bytes: hold
            used_bytes: used
          enableCache: true
          cacheExpire: 60s

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ tenant_id, __all_unit_config.name, max_cpu, min_cpu, max_memory, min_memory, max_iops, min_iops from __all_resource_pool, __all_unit_config, __all_unit  where __all_resource_pool.unit_config_id = __all_unit_config.unit_config_id and __all_unit.resource_pool_id = __all_resource_pool.resource_pool_id and __all_unit.svr_ip = ?
          params: [ ob_svr_ip ]
          sqlSlowThreshold: 100ms
          name: ob_unit_config
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
            unit_config_name: name
          metrics:
            max_cpu: max_cpu
            min_cpu: min_cpu
            max_memory_bytes: max_memory
            max_iops: max_iops
            min_iops: min_iops
            max_session_num: max_session_num
          enableCache: true
          cacheExpire: 60s
        - sql: select /* MONITOR_AGENT */ DBA_OB_RESOURCE_POOLS.tenant_id, DBA_OB_UNIT_CONFIGS.name, DBA_OB_UNIT_CONFIGS.max_cpu, DBA_OB_UNIT_CONFIGS.min_cpu, DBA_OB_UNIT_CONFIGS.memory_size as max_memory, DBA_OB_UNIT_CONFIGS.max_iops, DBA_OB_UNIT_CONFIGS.min_iops from DBA_OB_RESOURCE_POOLS, DBA_OB_UNIT_CONFIGS, DBA_OB_UNITS  where DBA_OB_RESOURCE_POOLS.unit_config_id = DBA_OB_UNIT_CONFIGS.unit_config_id and DBA_OB_UNITS.resource_pool_id = DBA_OB_RESOURCE_POOLS.resource_pool_id and DBA_OB_UNITS.svr_ip = ?
          params: [ob_svr_ip]
          sqlSlowThreshold: 100ms
          name: ob_unit_config
          minObVersion: 4.0.0.0
          maxObVersion: ~
          tags:
            ob_tenant_id: tenant_id
            unit_config_name: name
          metrics:
            max_cpu: max_cpu
            min_cpu: min_cpu
            max_memory_bytes: max_memory
            min_memory_bytes: min_memory
            max_iops: max_iops
            min_iops: min_iops
          enableCache: true
          cacheExpire: 60s

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) QUERY_TIMEOUT(100000000) */ tenant_id, 1 as role, case when cnt is null then 0 else cnt end as cnt from (select tenant_id, count(*) as cnt from __all_virtual_partition_info where svr_ip = ? and svr_port = ? group by tenant_id)
          params: [ ob_svr_ip, ob_svr_port ]
          sqlSlowThreshold: 100ms
          name: ob_partition
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            ob_tenant_id: tenant_id
            role: role
          metrics:
            num: cnt
          enableCache: true
          cacheExpire: 1h

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ zone, name, value, time_to_usec(now()) as current from __all_zone
          name: ob_zone
          sqlSlowThreshold: 100ms
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            zone: zone
            name: name
          condition: ob_is_rootservice
          metrics:
            stat: value
            current_timestamp: current
          enableCache: true
          cacheExpire: 60s

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ __all_tenant.tenant_name as tenant_name, cast(v_acc_response_time.response_time / 1000000 as float) as response_time_seconds,  v_acc_response_time.count as bucket, case when  v_acc_response_time.response_time = (select max(response_time) from __all_virtual_query_response_time) then v_acc_response_time.count else null end as count, case when  v_acc_response_time.response_time = (select max(response_time) from __all_virtual_query_response_time) then cast(v_acc_response_time.sum / 1000000 as float) else null end as sum from (select b.tenant_id, b.response_time  as response_time, sum(a.count) as count, sum(a.total) as sum from  __all_virtual_query_response_time a,  __all_virtual_query_response_time b where a.response_time <= b.response_time and a.svr_ip = b.svr_ip and a.svr_port = b.svr_port and b.svr_ip = ? and b.svr_port = ? group by b.tenant_id, b.response_time) v_acc_response_time,  __all_tenant where v_acc_response_time.tenant_id = __all_tenant.tenant_id;
          params: [ob_svr_ip, ob_svr_port]
          name: ob_query_response_time_seconds
          sqlSlowThreshold: 100ms
          minObVersion: ~
          maxObVersion: 4.0.0.0
          tags:
            tenant_name: tenant_name
            le: response_time_seconds
          metrics:
            bucket: bucket
            count: count
            sum: sum
          enableCache: true
          cacheExpire: 60s

        - sql: select /*+ MONITOR_AGENT READ_CONSISTENCY(WEAK) */ count(1) as cnt from __all_virtual_trans_stat where part_trans_action > 2 and ctx_create_time < date_sub(now(), interval 600 second) and svr_ip = ? and svr_port = ?
          params: [ob_svr_ip, ob_svr_port]
          name: ob_trans
          sqlSlowThreshold: 100ms
          condition: ob_is_rootservice
          minObVersion: ~
          maxObVersion: 4.0.0.0
          metrics:
            expire_num: cnt
          enableCache: true
          cacheExpire: 60s

jointableProcessor: &jointableProcessor
  plugin: jointable
  config:
    timeout: 10s
    pluginConfig:
      connection:
        url: ${monagent.ob.monitor.user}:${monagent.ob.monitor.password}@tcp(127.0.0.1:${monagent.ob.sql.port})/oceanbase?interpolateParams=true
        maxIdle: 2
        maxOpen: 32
      joinTableConfigs:
        - queryDataConfigs:
            - querySql: SELECT tenant_id ob_tenant_id, tenant_name FROM __all_tenant
              queryArgs: [ ]
              minOBVersion: ~
              maxOBVersion: 4.0.0.0
          cacheExpire: 3m
          conditions:
            - metrics: [ "ob_cache", "ob_partition", "ob_table", "ob_plan_cache", "ob_waitevent", "ob_system_event", "ob_tenant_task", "ob_clog", "ob_memstore", "ob_compaction", "ob_unit_config", "ob_tenant", "ob_tenant_server", "ob_tenant_server_available" ]
              tagNames: [ "ob_tenant_id" ]
              removeNotMatchedTagValueMessage: true
        - queryDataConfigs:
            - querySql: SELECT tenant_id ob_tenant_id, tenant_name FROM DBA_OB_TENANTS WHERE tenant_type<>'META'
              queryArgs: [ ]
              minOBVersion: 4.0.0.0
              maxOBVersion: ~
          cacheExpire: 3m
          conditions:
            - metrics: [ "ob_cache", "ob_partition", "ob_table", "ob_plan_cache", "ob_waitevent", "ob_system_event", "ob_tenant_task", "ob_clog", "ob_memstore", "ob_compaction", "ob_unit_config" ]
              tagNames: [ "ob_tenant_id" ]
              removeNotMatchedTagValueMessage: true
        - queryDataConfigs:
            - querySql: SELECT tenant_id ob_tenant_id, tenant_name FROM __all_tenant
              queryArgs: [ ]
              minOBVersion: ~
              maxOBVersion: 4.0.0.0
          cacheExpire: 3m
          conditions:
            - metrics: [ "ob_sysstat" ]
              tagNames: [ "ob_tenant_id" ]
              removeNotMatchedTagValueMessage: false # contain internal tenant, do not remove
        - queryDataConfigs:
            - querySql: SELECT tenant_id ob_tenant_id, tenant_name FROM DBA_OB_TENANTS WHERE tenant_type<>'META'
              queryArgs: [ ]
              minOBVersion: 4.0.0.0
              maxOBVersion: ~
          cacheExpire: 3m
          conditions:
            - metrics: [ "ob_sysstat" ]
              tagNames: [ "ob_tenant_id" ]
              removeNotMatchedTagValueMessage: false # contain internal tenant, do not remove

obRetagProcessor: &obRetagProcessor
  plugin: retagProcessor
  config:
    timeout: 10s
    pluginConfig:
      newTags:
        app: OB
        obzone: ${monagent.ob.zone.name}
        svr_ip: ${monagent.host.ip}
        svr_port: ${monagent.ob.rpc.port}
        ob_cluster_name: ${monagent.ob.cluster.name}
        ob_cluster_id: ${monagent.ob.cluster.id}

basicExporter: &basicExporter
  plugin: prometheusExporter
  config:
    timeout: 10s
    pluginConfig:
      formatType: fmtText
      exposeUrl: /metrics/ob/basic

extraExporter: &extraExporter
  plugin: prometheusExporter
  config:
    timeout: 10s
    pluginConfig:
      formatType: fmtText
      exposeUrl: /metrics/ob/extra

modules:
  - module: monitor.ob
    moduleType: monagent.pipeline
    process: ob_monagent
    config:
      name: monitor.ob
      status: ${monagent.pipeline.ob.status}
      pipelines:
        - name: ob_basic
          config:
            scheduleStrategy: bySource
          structure:
            inputs:
              - <<: *obInputBasic
            processors:
              - <<: *obRetagProcessor
              - <<: *jointableProcessor
            exporter:
              <<: *basicExporter
        - name: ob_extra
          config:
            scheduleStrategy: bySource
          structure:
            inputs:
              - <<: *obInputExtra
            processors:
              - <<: *obRetagProcessor
              - <<: *jointableProcessor
            exporter:
              <<: *extraExporter