This topic describes the tables that comprise the ClustrixDB system database.
System tables cannot be altered by users. The following is informational only.
The alter_progress table provides a rough estimate of the progress for an ALTER.
Name | Description |
---|---|
db | The ID for the session. |
src_relation_name | The name of the source relation. |
start | Timestamp for when the alter started. |
rows_read | The number of rows read by the ALTER. |
src_relation_est_rows | The estimated number of rows in the source relation |
completion_est | A rough estimate of when the ALTER might finish. |
The databases table contains one row for each database in the system. Its columns are as follows:
Name | Description |
---|---|
db | Contains an ID for each database. It corresponds to the db column in the relations table. |
user | Contains the ID of the user who created the database. It corresponds to the user column of the users table. |
name | Name of the database. |
hidden | Indicates whether the table is hidden from users when performing queries. 1 = yes, 0 = no. |
cscl | An internal value that encodes the character set and collation sequence utilized by a given database. To translate this encoded value to English, a special function is available as follows: select OIDTYPE_CSCL_PROP(cscl, 'charset') from system.databases where name = 'sample'; select OIDTYPE_CSCL_PROP(cscl, 'collation') from system.databases where name = 'sample'; |
The cluster_session_stats table contains per-session statistics about transactions and statement types.
This table contains a row for each open transaction.
The mysql_master_status table stores information about each replication slave that is currently connected to the cluster.
The cluster performs ongoing optimization tasks in background mode. In rare cases, this background activity might affect query performance. In such cases, the Rebalancer, which moves data between disks and nodes to ensure even distribution, is a likely source. Rebalancing typically runs after an import, and occasionally runs as disk utilization increases. Because Rebalancer sessions can be brief, the rebalancer_activity_log table is a better source for detailed information.
Values returned in the action column are as follows:
Action | Description |
---|---|
split | A slice has grown too large and is being split into two smaller slices. |
move | A slice is being moved to another disk or node to optimize the load. |
copy | A slice is being duplicated, usually during the reprotection stage after a hardware failure. |
This table contains a detailed log of Rebalancer actions and the reasons for those actions. See Monitoring Data Rebalancing Activity.
The relations table contains one row for each table in the system.
The representations table contains one row per representation in the system. Each relation and each index on that relation constitutes a representation. The base column indicates when the representation is the Base Representation.
The sessions table contains information about all currently connected and recently disconnected client sessions.
The stats table contains internally-collected statistics about database and cluster performance. Most of these statistics are for internal use, but some are of general interest.
The table_replicas table contains one row for each replica in the system and records each replica's size and node. This table can be used to determine the total calculated size of an index and to find poorly-distributed indexes. Each slice is replicated two or more times and each replica resides on a different node. Each replica of a slice contains the same data.
The table_sizes table records the total size of each table in bytes, including its base representation and all of its indexes.
Representations are split into slices and distributed throughout your cluster. The table_slices table enumerates those slices and indexes and contains one row for each slice in each table.
Each slice contains a unique set of rows from the table. The complete table is represented by assembling all the rows from the table slices. The rows are allocated to slices using a hash that is based on the row's index or indices. The number of slices per table depends on many factors including the size of the cluster. To configure this setting, issue the ALTER TABLE SET SLICES = <N> command. Changing the slice count can improve performance. For details, contact Clustrix Support.
To display the names of the database, table, and indexes for each slice, join on the slice column.
The version_history table provides a history of installation and upgrade versions.
Database user, host, and password combinations are stored in the users table. Join to this table to display user names instead of user ids. See also Managing Users and Privileges and Migrating Users and Permissions.