Count with tag filter returns the unfiltered number of rows

[akhilman]

What type of bug is this?

Incorrect result

What subsystems are affected?

Standalone mode

Minimal reproduce step

drop table if exists count_where_bug;
create table count_where_bug (
    tag String,
    ts TimestampMillisecond time index,
    num Int64,
    primary key (tag),
) engine=mito with('append_mode'='true');

insert into count_where_bug (tag, ts, num)
values  ('a', '2024-09-06T06:00:01Z', 1),
        ('a', '2024-09-06T06:00:02Z', 2),
        ('a', '2024-09-06T06:00:03Z', 3),
        ('b', '2024-09-06T06:00:04Z', 4),
        ('b', '2024-09-06T06:00:05Z', 5);

select count(1) from count_where_bug where tag = 'b';

What did you expect to see?

I expect to get 2 as result.

What did you see instead?

But I am getting total number of rows - 5.

What operating system did you use?

ArchLinux x64

What version of GreptimeDB did you use?

0.9.2 from docker.io

Relevant log output and stack trace

2024-09-06T13:32:32.355432Z  INFO LocalManager::submit_root_procedure: common_procedure::local::runner: Runner metasrv-procedure::DropTable-5e18069e-21d2-47ba-b186-257e7f20887e starts
2024-09-06T13:32:32.355587Z  INFO LocalManager::submit_root_procedure: common_meta::ddl::drop_table: Deleted table metadata for table 1083
2024-09-06T13:32:32.355732Z  INFO mito2::worker::handle_drop: Try to drop region: 4651449581568(1083, 0), worker: 3
2024-09-06T13:32:32.360541Z  INFO mito2::region: Stopped region manifest manager, region_id: 4651449581568(1083, 0)
2024-09-06T13:32:32.360602Z  INFO log_store::raft_engine::log_store: Namespace 4651449581568 obsoleted 1 entries, compacted index: 1, span: (None, None)
2024-09-06T13:32:32.360633Z  INFO mito2::worker::handle_drop: Region 4651449581568(1083, 0) is dropped logically, but some files are not deleted yet
2024-09-06T13:32:32.360678Z  INFO LocalManager::submit_root_procedure:RegionInvoker::handle_region_request:RegionServer::handle_region_request{region_id="4651449581568(1083, 0)"}:handle_request{request_type="Drop"}: datanode::region_server: Region 4651449581568(1083, 0) is deregistered from engine mito
2024-09-06T13:32:32.360808Z  INFO LocalManager::submit_root_procedure: common_procedure::local::runner: Procedure metasrv-procedure::DropTable-5e18069e-21d2-47ba-b186-257e7f20887e done
2024-09-06T13:32:32.360880Z  INFO LocalManager::submit_root_procedure: common_procedure::local::runner: Runner metasrv-procedure::DropTable-5e18069e-21d2-47ba-b186-257e7f20887e exits
2024-09-06T13:32:32.360912Z  INFO on_query{protocol="mysql"}:mysql::do_query:do_query:execute_stmt:drop_tables:DdlManager::submit_ddl_task: common_meta::ddl_manager: Table: 1083 is dropped via procedure_id ProcedureId(5e18069e-21d2-47ba-b186-257e7f20887e)
2024-09-06T13:32:32.361322Z  INFO LocalManager::submit_root_procedure: common_procedure::local::runner: Runner metasrv-procedure::CreateTable-965e4502-5989-4e82-901e-aa9cf6f10bf7 starts
2024-09-06T13:32:32.361584Z  INFO mito2::manifest::manager: Checkpoint not found in data/greptime/public/1084/1084_0000000000/manifest/, build manifest from scratch
2024-09-06T13:32:32.361664Z  INFO mito2::manifest::manager: Creating region manifest in data/greptime/public/1084/1084_0000000000/manifest/ with metadata RegionMetadata { column_metadatas: [[tag String null Tag 0], [ts TimestampMillisecond not null time_index metadata={"greptime:time_index": "true"} Timestamp 1], [num Int64 null Field 2]], time_index: 1, primary_key: [0], region_id: 4655744548864(1084, 0), schema_version: 0 }
2024-09-06T13:32:32.367128Z  INFO mito2::worker::handle_create: A new region created, worker: 0, region: RegionMetadata { column_metadatas: [[tag String null Tag 0], [ts TimestampMillisecond not null time_index metadata={"greptime:time_index": "true"} Timestamp 1], [num Int64 null Field 2]], time_index: 1, primary_key: [0], region_id: 4655744548864(1084, 0), schema_version: 0 }
2024-09-06T13:32:32.367195Z  INFO LocalManager::submit_root_procedure:RegionInvoker::handle_region_request:RegionServer::handle_region_request{region_id="4655744548864(1084, 0)"}:handle_request{request_type="Create"}: datanode::region_server: Region 4655744548864(1084, 0) is registered to engine mito
2024-09-06T13:32:32.367364Z  INFO LocalManager::submit_root_procedure: common_meta::ddl::create_table: Created table metadata for table 1084
2024-09-06T13:32:32.367389Z  INFO LocalManager::submit_root_procedure: common_procedure::local::runner: Procedure metasrv-procedure::CreateTable-965e4502-5989-4e82-901e-aa9cf6f10bf7 done
2024-09-06T13:32:32.367442Z  INFO LocalManager::submit_root_procedure: common_procedure::local::runner: Runner metasrv-procedure::CreateTable-965e4502-5989-4e82-901e-aa9cf6f10bf7 exits
2024-09-06T13:32:32.367479Z  INFO on_query{protocol="mysql"}:mysql::do_query:do_query:execute_stmt:create_table:create_table_inner:DdlManager::submit_ddl_task: common_meta::ddl_manager: Table: 1084 is created via procedure_id ProcedureId(965e4502-5989-4e82-901e-aa9cf6f10bf7)
2024-09-06T13:32:32.367497Z  INFO on_query{protocol="mysql"}:mysql::do_query:do_query:execute_stmt:create_table:create_table_inner: operator::statement::ddl: Successfully created table 'greptime.public.count_where_bug' with table id 1084
.

[v0y4g3r]

This issue may be related to #4668 , please try v0.9.3

[akhilman]

Issue still present in v0.9.3 from docker.io

[evenyag]

This is a regression after #4545 and #4564. It should affect the latest version.

[evenyag]

We'll fix it and release the next version as soon as possible. Thanks for reporting this issue! We need to add a sqlness for this case.

It should only affect count() under append-only table.

[evenyag]

https://github.com/GreptimeTeam/greptimedb/releases/tag/v0.10.0-nightly-20240909 already fixes this.