【 使用环境 】测试环境
【 OB or 其他组件 】OCP
【 使用版本 】 v4.2.2
【问题描述】ocp_monitor sql占用CPU过高. 有什么参数可以降低这个sql的占用/或者执行频率吗 ?
- 被管理集群为 observer 4.3.5.5, 1-1-1 集群, 8C 64G .
- ocp_monitor 租户配置为 4C 12G
- ocp_monitor TOP 1 SQL
select
/*+
index(ob_hist_plan_cache_plan_logic_explain idx_ob_db_id_sql_id_plan_union_hash)
query_timeout(30000000) read_consistency(weak)
*/
collect_time as collectTimeUs,
ob_cluster_id as obClusterId,
cluster_name as clusterName,
ob_tenant_id as obTenantId,
ob_server_id as obServerId,
plan_id as planId,
first_load_time as firstLoadTimeUs,
id as id,
operator as operator,
IF(object_type = ?, object_alias, ?) as objectName,
IF(object_type = ?, object_name, ?) as objectAlias,
cardinality as rows,
cost as cost,
real_cardinality as realRows,
real_cost as realCost,
io_cost as ioCost,
cpu_cost as cpuCost,
projection as projection,
filter_predicates as filterPredicates,
startup_predicates as startupPredicates,
rowset as rowset,
access_predicates as accessPredicates,
partition_start as partitionStart,
special_predicates as specialPredicates,
depth as depth,
other_tag as otherTag,
remarks as remarks,
other_xml as otherXml,
optimizer as optimizer,
other as other,
gmt_create as create_time
from
ob_hist_plan_cache_plan_logic_explain
where
collect_time >= ?
and collect_time < ?
and ob_cluster_id = ?
and cluster_name = ?
and ob_tenant_id = ?
and ob_server_id = ?
and plan_union_hash = ?
【附件及日志】
【备注】基于 LLM 和开源文档 RAG 的论坛小助手已开放测试,在发帖时输入 [@论坛小助手] 即可召唤小助手,欢迎试用!

