recover shard

Executes all DDL statements on the specified shard (database), starting from the one that was previously executed with errors. The command is intended to perform all skipped DDL changes after database administrator fixes shard issues.

Syntax

recover shard -gdspool pool
              -shard shard_name
             [-skip_first|-ignore_first]
             [-full]

Options

Table 2-68 GDSCTL recover shard Options

Option	Description
-full	Full recovery mode.
-gdspool pool	Specify the GDS pool. If not specified and there is only one GDS pool with access granted to user, it will be used by default.
-ignore_first	Make first failed DDL statement obsolete.
-shard shard_name	The name of the shard.
-skip_first	Skip the first failed DDL statement.

Usage Notes

Use SKIP_FIRST to skip first DDL. This is typically required after manual fix done by database administrator. For example, if CREATE TABLE statement fails because of a lack of space, the database administrator fixes the issue and re-executes CREATE TABLE. To avoid ORA-39151 (table exists) in RECOVER SHARD the database administrator must specify -SKIP_FIRST.

Use IGNORE_FIRST to mark the first DDL as obsolete. This is required when the wrong DDL statement was specified and failed on all shards. In this case, you need to mark it down as obsolete. FULL mode performs a complete recovery, including DDL operations, failed chunk migration, tablespace sets reconstruction, and database parameters.

Examples

Recover shard shd1.

GDSCTL> recover shard -shard shd1