查询按时间过滤遇到的诡异问题 #10754
Replies: 1 comment 3 replies
-
|
这个环境还在吗？日志之类的能提供下不可以协助分析下。 |
Beta Was this translation helpful? Give feedback.
-
|
日志如下: java.util.concurrent.TimeoutException: Waited 299988 milliseconds (plus 67062 nanoseconds delay) for io.grpc.stub.ClientCalls$GrpcFuture@45dc7e6d[status=PENDING, info=[GrpcFuture{clientCall=ClientCallImpl{method=MethodDescriptor{fullMethodName=doris.PBackendService/fetch_data, type=UNARY, idempotent=false, safe=false, sampledToLocalTracing=true, requestMarshaller=io.grpc.protobuf.lite.ProtoLiteUtils$MessageMarshaller@1cdaac78, responseMarshaller=io.grpc.protobuf.lite.ProtoLiteUtils$MessageMarshaller@2d1e1589, schemaDescriptor=org.apache.doris.proto.PBackendServiceGrpc$PBackendServiceMethodDescriptorSupplier@55cf856b}}}]] |
Beta Was this translation helpful? Give feedback.
-
|
这个可能是生成的解析规划就不一样。能explain下两种情况下的执行计划不？ |
Beta Was this translation helpful? Give feedback.
-
因为一直没有头绪，就检查了一下环境。 |
Beta Was this translation helpful? Give feedback.
-
某一个表test，按datetieme格式的字段ctime进行分区，分区为天。
create table
test(user_id int,
ctime datetime,
user_name text,
field1 xxx,
field2 xxx,
....
fieldN xxx
) engine=olap
DUPLICATE KEY(
user_id,ctime)PARTITION BY RANGE(
ctime)DISTRUBUTED BY HASH('user_id`) BUCKET10
PROPERTIES (
"dynamic_partition.enable" = "true",
"dynamic_partition.time_unic" = "DAY",
.....
);
在积累到百亿级别的数据的时候，按时间进行查询，出现了timeout:
select * from test where ctime > '2022-07-01 00:00:00' and ctime < '2022-07-02 00:00:00';但是如果把时间稍微调整一下,加一秒，数据很快就返回了：
select * from test where ctime > '2022-07-01 00:00:01' and ctime < '2022-07-02 00:00:01';查看了query profile，日志显示完成查询只花了5秒多，但是客户端却没有反映，一直到出现time out错误。
Beta Was this translation helpful? Give feedback.
All reactions