distribution: local
vectorized: true

• root
│
├── • insert
│   │ into: t2(id, c, d, grp, v)
│   │
│   └── • scan buffer
│         estimated row count: 100
│         label: buffer 1
│
└── • subquery
    │ id: @S1
    │ original sql: DELETE FROM t WHERE c < 10 RETURNING *
    │ exec mode: discard all rows
    │
    └── • buffer
        │ label: buffer 1
        │
        └── • delete
            │ estimated row count: 100
            │ from: t
            │
            └── • index join
                │ estimated row count: 100
                │ table: t@t_pkey
                │ locking strength: for update
                │
                └── • scan
                      estimated row count: 100 (10% of the table; stats collected 40 seconds ago)
                      table: t@idx_c
                      spans: [ - /9]
                      locking strength: for update

index recommendations: 1
1. type: index replacement
   SQL commands: CREATE INDEX ON capture.public.t (c) STORING (d, grp, v) /* After successfully creating the replacement index, manually run: `ALTER INDEX capture.public.t@idx_c NOT VISIBLE` and then, after verifying workload performance, manually run: `DROP INDEX capture.public.t@idx_c` */;
