feat(fuzz): add create logical table target

[CookiePieWw]

I hereby agree to the terms of the GreptimeDB CLA.

Refer to a related PR or issue link (optional)

#3741 and #3174

What's changed and what's your intention?

Fuzz test for metric engine. Create some logical tables on the same physical table.

Checklist

• I have written the necessary rustdoc comments.
• I have added the necessary unit tests and integration tests.
• This PR does not require documentation updates.

[WenyXu]

We can start by mocking the usage of GreptimeDB as Prometheu's backend(https://docs.greptime.com/user-guide/write-data/prometheus)

We always create two physical table with two columns: a timestamp(millisecond, datatypes) column and a value(double) column. The datatype is following the Prometheus document).

i.e.,

CREATE TABLE phy (ts timestamp time index, val double) engine=metric with ("physical_metric_table" = "");

Then we will get a physical table.

select column_name, data_type, semantic_type  from information_schema.columns where table_name = 'phy';
+-------------+--------------+---------------+
| column_name | data_type    | semantic_type |
+-------------+--------------+---------------+
| ts          | timestamp(3) | TIMESTAMP     |
| val         | double       | FIELD         |
+-------------+--------------+---------------+

Then, the fuzz target randomly creates many logical tables and validates the datatypes of the logical and physical table.

e.g.,

CREATE TABLE t2 (ts timestamp time index, val double, host string, idc string, primary key(idc, host)) engine = metric with ("on_physical_table" = "phy");

Validating the datatypes of the physical table.

select column_name, data_type, semantic_type, column_default from information_schema.columns where table_name = 'phy'; 
+-------------+-----------------+---------------+----------------+
| column_name | data_type       | semantic_type | column_default |
+-------------+-----------------+---------------+----------------+
| ts          | timestamp(3)    | FIELD         | NULL           |
| val         | double          | FIELD         | NULL           |
| __table_id  | int unsigned    | TAG           | NULL           |
| __tsid      | bigint unsigned | TAG           | NULL           |
| host        | string          | TAG           | NULL           |
| idc         | string          | TAG           | NULL           |
+-------------+-----------------+---------------+----------------+

Validating the datatypes of the logical table.

select column_name, data_type, semantic_type, column_default from information_schema.columns where table_name = 't2';
+-------------+--------------+---------------+----------------+
| column_name | data_type    | semantic_type | column_default |
+-------------+--------------+---------------+----------------+
| host        | string       | TAG           | NULL           |
| idc         | string       | TAG           | NULL           |
| ts          | timestamp(3) | TIMESTAMP     | NULL           |
| val         | double       | FIELD         | NULL           |
+-------------+--------------+---------------+----------------+

[WenyXu]

We can start by mocking the usage of GreptimeDB as Prometheu's backend(https://docs.greptime.com/user-guide/write-data/prometheus)

We always create two physical table with two columns: a timestamp(millisecond, datatypes) column and a value(double) column. The datatype is following the Prometheus document).

i.e.,

CREATE TABLE phy (ts timestamp time index, val double) engine=metric with ("physical_metric_table" = "");

Then we will get a physical table.

select column_name, data_type, semantic_type  from information_schema.columns where table_name = 'phy';
+-------------+--------------+---------------+
| column_name | data_type    | semantic_type |
+-------------+--------------+---------------+
| ts          | timestamp(3) | TIMESTAMP     |
| val         | double       | FIELD         |
+-------------+--------------+---------------+

Then, the fuzz target randomly creates many logical tables and validates the datatypes of the logical and physical table.

e.g.,

CREATE TABLE t2 (ts timestamp time index, val double, host string, idc string, primary key(idc, host)) engine = metric with ("on_physical_table" = "phy");

Validating the datatypes of the physical table.

select column_name, data_type, semantic_type, column_default from information_schema.columns where table_name = 'phy'; 
+-------------+-----------------+---------------+----------------+
| column_name | data_type       | semantic_type | column_default |
+-------------+-----------------+---------------+----------------+
| ts          | timestamp(3)    | FIELD         | NULL           |
| val         | double          | FIELD         | NULL           |
| __table_id  | int unsigned    | TAG           | NULL           |
| __tsid      | bigint unsigned | TAG           | NULL           |
| host        | string          | TAG           | NULL           |
| idc         | string          | TAG           | NULL           |
+-------------+-----------------+---------------+----------------+

Validating the datatypes of the logical table.

select column_name, data_type, semantic_type, column_default from information_schema.columns where table_name = 't2';
+-------------+--------------+---------------+----------------+
| column_name | data_type    | semantic_type | column_default |
+-------------+--------------+---------------+----------------+
| host        | string       | TAG           | NULL           |
| idc         | string       | TAG           | NULL           |
| ts          | timestamp(3) | TIMESTAMP     | NULL           |
| val         | double       | FIELD         | NULL           |
+-------------+--------------+---------------+----------------+

BTW, there is the bug. The semantic_type in the 1st row should be TIMESTAMP.

select column_name, data_type, semantic_type, column_default from information_schema.columns where table_name = 'phy'; 
+-------------+-----------------+---------------+----------------+
| column_name | data_type       | semantic_type | column_default |
+-------------+-----------------+---------------+----------------+
| ts          | timestamp(3)    | FIELD         | NULL           |
| val         | double          | FIELD         | NULL           |
| __table_id  | int unsigned    | TAG           | NULL           |
| __tsid      | bigint unsigned | TAG           | NULL           |
| host        | string          | TAG           | NULL           |
| idc         | string          | TAG           | NULL           |
+-------------+-----------------+---------------+----------------+

[codecov]

Codecov Report

Attention: Patch coverage is 98.18182% with 3 lines in your changes are missing coverage. Please review.

Project coverage is 85.32%. Comparing base (bf21527) to head (f569e32).
Report is 26 commits behind head on main.

❗ Current head f569e32 differs from pull request most recent head 6ca29f5. Consider uploading reports for the commit 6ca29f5 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3756      +/-   ##
==========================================
- Coverage   85.53%   85.32%   -0.22%     
==========================================
  Files         946      953       +7     
  Lines      159564   163046    +3482     
==========================================
+ Hits       136489   139120    +2631     
- Misses      23075    23926     +851     

[WenyXu]

The tests of #3772 may help you to build the fuzz test

[WenyXu]

LGTM

[fengjiachun]

LGTM