Show ranges in explain indexes

[cwurm]

When investigating query performance, it can be very useful to see how many ranges ClickHouse will read from. Today, this is shown in a log line in system.text_log, but we can have it earlier, before executing the query in EXPLAIN indexes = 1.

For example, it is useful for a case like this where the number of granules to be read is almost the same, but the number of ranges is vastly different:

create table t1 (x int, y int, z int) order by (x, y);
create table t2 (x int, y int, z int) order by (y, x);

insert into t1 select number % 100, number % 100, number from numbers(100_000_000);
insert into t2 select * from t1;

optimize table t1 final;
optimize table t2 final;

explain indexes = 1 select avg(z) from t1 where y in (select number * 10 from numbers(10));

    ┌─explain────────────────────────────────────────────────┐
 1. │ CreatingSets (Create sets before main query execution) │
 2. │   Expression ((Project names + Projection))            │
 3. │     Aggregating                                        │
 4. │       Expression (Before GROUP BY)                     │
 5. │         Expression                                     │
 6. │           ReadFromMergeTree (default.t1)               │
 7. │           Indexes:                                     │
 8. │             PrimaryKey                                 │
 9. │               Keys:                                    │
10. │                 y                                      │
11. │               Condition: (y in 10-element set)         │
12. │               Parts: 1/1                               │
13. │               Granules: 1310/12208                     │
14. │               Ranges: 90                               │
    └────────────────────────────────────────────────────────┘
    
explain indexes = 1 select avg(z) from t2 where y in (select number * 10 from numbers(10));

    ┌─explain────────────────────────────────────────────────┐
 1. │ CreatingSets (Create sets before main query execution) │
 2. │   Expression ((Project names + Projection))            │
 3. │     Aggregating                                        │
 4. │       Expression (Before GROUP BY)                     │
 5. │         Expression                                     │
 6. │           ReadFromMergeTree (default.t2)               │
 7. │           Indexes:                                     │
 8. │             PrimaryKey                                 │
 9. │               Keys:                                    │
10. │                 y                                      │
11. │               Condition: (y in 10-element set)         │
12. │               Parts: 1/1                               │
13. │               Granules: 1230/12208                     │
14. │               Ranges: 10                               │
    └────────────────────────────────────────────────────────┘

Changelog category (leave one):

• Improvement

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):

Show the number of ranges to be read in the output of EXPLAIN indexes = 1.

[clickhouse-gh]

Workflow [PR], commit [c651977]

[cwurm]

Interestingly, it can show fewer ranges than expected (like 1 range for 2 parts below) because of things like the query condition cache. I need to think about this a bit.

    ┌─explain─────────────────────────────────────────────────┐
 1. │ Expression ((Project names + Projection))               │
 2. │   Aggregating                                           │
 3. │     Expression (Before GROUP BY)                        │
 4. │       Expression                                        │
 5. │         ReadFromMergeTree (default.t_skip_index_insert) │
 6. │         Indexes:                                        │
 7. │           Skip                                          │
 8. │             Name: idx_a                                 │
 9. │             Description: minmax GRANULARITY 1           │
10. │             Parts: 2/2                                  │
11. │             Granules: 50/50                             │
12. │           Skip                                          │
13. │             Name: idx_b                                 │
14. │             Description: set GRANULARITY 1              │
15. │             Parts: 2/2                                  │
16. │             Granules: 50/50                             │
17. │             Ranges: 1                                   │
    └─────────────────────────────────────────────────────────┘

[nickitat]

because of things like the query condition cache

they could be disabled.
pls update reference files.

[clickhouse-gh]

Dear @nickitat, this PR hasn't been updated for a while. You will be unassigned. Will you continue working on it? If so, please feel free to reassign yourself.

[clickhouse-gh]

Workflow [PR], commit [9a6a3d9]

Summary: ❌

job_name	test_name	status	info	comment
Stateless tests (amd_debug, distributed plan, s3 storage)		failure
	03352_concurrent_rename_alter	FAIL

[cwurm]

@nickitat I finally found time to fix all the test outputs. The failure in 03352_concurrent_rename_alter looks unrelated.

[rschu1ze]

Gentlemen, please don't forget to update the docs. Did that here.