doris执行select count查询特别慢

Viewed 57

doris版本2.1.3
执行select count(*) from table耗时很久,之前查询耗时在3秒左右,总量数据约23亿,当前查询时间需要
456214 ms,多次执行,耗时都特别高,Fe,BE节点资源占用不高,请问有大佬你知道是什么情况吗

2 Answers

看看表结构以及查询对应的profile

表结构信息:ENGINE=OLAP UNIQUE KEY(province_name, sms_port, sub_sms_port) COMMENT '子端口全量数据登记表' PARTITION BY LIST(province_name) (PARTITION Shanghai VALUES IN ("上海"), PARTITION Yunnan VALUES IN ("云南"), PARTITION Inner_Mongolia VALUES IN ("内蒙古"), PARTITION Beijing VALUES IN ("北京"), PARTITION Jilin VALUES IN ("吉林"), PARTITION Sichuan VALUES IN ("四川"), PARTITION Tianjin VALUES IN ("天津"), PARTITION Ningxia VALUES IN ("宁夏"), PARTITION Anhui VALUES IN ("安徽"), PARTITION Shandong VALUES IN ("山东"), PARTITION Shanxi VALUES IN ("山西"), PARTITION Guangdong VALUES IN ("广东"), PARTITION Guangxi VALUES IN ("广西"), PARTITION Xinjiang VALUES IN ("新疆"), PARTITION Jiangsu VALUES IN ("江苏"), PARTITION Jiangxi VALUES IN ("江西"), PARTITION Hebei VALUES IN ("河北"), PARTITION Henan VALUES IN ("河南"), PARTITION Zhejiang VALUES IN ("浙江"), PARTITION Hainan VALUES IN ("海南"), PARTITION Hubei VALUES IN ("湖北"), PARTITION Hunan VALUES IN ("湖南"), PARTITION Gansu VALUES IN ("甘肃"), PARTITION Fujian VALUES IN ("福建"), PARTITION Tibet VALUES IN ("西藏"), PARTITION Guizhou VALUES IN ("贵州"), PARTITION Liaoning VALUES IN ("辽宁"), PARTITION ChongqingAnhui VALUES IN ("重庆"), PARTITION Shaanxi VALUES IN ("陕西"), PARTITION Qinhai VALUES IN ("青海"), PARTITION Heilongjiang VALUES IN ("黑龙江")) DISTRIBUTED BY HASH(sub_sms_port) BUCKETS AUTO PROPERTIES ( "replication_allocation" = "tag.location.default: 3", "min_load_replica_num" = "-1", "is_being_synced" = "false", "storage_medium" = "hdd", "storage_format" = "V2", "inverted_index_storage_format" = "V1", "enable_unique_key_merge_on_write" = "true", "light_schema_change" = "true", "disable_auto_compaction" = "false", "enable_single_replica_compaction" = "false", "group_commit_interval_ms" = "10000", "group_commit_data_bytes" = "134217728" );profile这个没有配置,只有单独的select count(*)