tidb_store_limit INSTANCE scope is broken

[morgo]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

Deploy a cluster which uses tikv and PD (mocktikv without PD won't work).

SET tidb_store_limit = 5; SELECT @@tidb_store_limit; SELECT SLEEP(31); SELECT @@tidb_store_limit;

2. What did you expect to see? (Required)

The value of tidb_store_limit should persist.

3. What did you see instead (Required)

It gets overwritten by the global value every 30 seconds because of the sysvar cache refreshing. This is because there is currently no good way to do both instance scope and global scope in the server.

tidb> SET tidb_store_limit = 5; SELECT @@tidb_store_limit; SELECT SLEEP(31); SELECT @@tidb_store_limit;
Query OK, 0 rows affected (0.00 sec)

+--------------------+
| @@tidb_store_limit |
+--------------------+
| 5                  |
+--------------------+
1 row in set (0.00 sec)

+-----------+
| SLEEP(31) |
+-----------+
|         0 |
+-----------+
1 row in set (31.00 sec)

+--------------------+
| @@tidb_store_limit |
+--------------------+
| 0                  |
+--------------------+
1 row in set (0.00 sec)

Because it is broken I suggest we remove the INSTANCE (session) scope. If all sysvars are either GLOBAL or INSTANCE, then we can implement an easier design for proper INSTANCE scope.

4. What is your TiDB version? (Required)

master

[morgo]

Introduced in #26649 (although previously was likely broken as well)

[github-actions]

Please check whether the issue should be labeled with 'affects-x.y' or 'fixes-x.y.z', and then remove 'needs-more-info' label.