OceanBase Database Enterprise Edition V2.2.77

V2.2.77

Version information

• Release date: May 8, 2021

• Version: V2.2.77

Overview

OceanBase Database V2.2.X is a generally available series that provides products of higher availability, compatibility, and security for customers to batch launch multiple projects.

Enhanced features

• Primary and standby clusters are enhanced.

• Physical backup and restore are supported.

• The encryption of an individual log-only replica at the zone level is supported.

• Template-based and non-template-based subpartitioned tables are supported.

• Drop and Truncate operations on partitioned tables are supported.

Fixed issues

• An error is returned for the aligned read of the 4 KB clog offset.

• An error is returned during the type derivation of an expression that contains NULL.

V2.2.77 BP1

Version information

• Release date: May 27, 2021.

• Version: V2.2.77 BP1

Enhanced features

You can specify the maximum memory usage of partitions in a transaction by setting the tenant-level parameter _tenant_max_trx_size.

Fixed issues

• The observer process unexpectedly exits when the memory is insufficient and the memory request fails.

• The plan cache has memory leaks.

• The Nested Loop Join operator may cause the observer process to unexpectedly exit in the TABLE ACCESS BY INDEX PRIMARY KEY operation.

• When foreign keys exist, the Semi-Join rewrite generates incorrect results.

• The observer process unexpectedly exits when an intermediate result set is released.

• In Oracle mode, the WIDTH_BUCKET() function is incompatible with Oracle when min_value is less than expression.

• In Oracle mode, the RPAD() function is incompatible with Oracle.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP2

Version information

• Release date: June 4, 2021

• Version: V2.2.77 BP2

Enhanced features

• In MySQL mode, using escape characters such as single quotation marks (' ') in a LIKE statement is supported.

• You can explicitly disable the enforced routing of a transaction after it is complete by setting the tenant-level parameter ob_proxy_readonly_transaction_routing_policy. This way, all SQL statements in the transaction are forwarded to the same OBServer node.

• The memory size that can be utilized by sql_audit of an OceanBase Database tenant is increased.

• The timeout wait for asynchronous I/O is optimized.

Fixed issues

• The RootServer node of the standby cluster occasionally exits unexpectedly.

• The parameter that specifies the maximum memory available for sql_audit does not take effect.

• When the GBK character set is used, two consecutive queries that use Chinese characters return incorrect filed name even if the data is correct.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP3

Version information

• Release date: June 18, 2021

• Version: V2.2.77 BP3

Enhanced features

• The LOG(), LN(), PI(), RADIANS(), ATAN(), ASIN(), ACOS(), and COT() functions are supported in MySQL mode.

• The MONTHNAME() function is supported in MySQL mode.

• The LAST_DAY() function is supported in MySQL mode.

• The QUARTER() function is supported in MySQL mode.

• The BIT_LENGTH() function is supported in MySQL mode.

Fixed issues

• The observer process unexpectedly exits during the packet return process of a query because of network exceptions.

• The parallel execution framework has memory leaks.

• The observer process unexpectedly exits during the process of obtaining the partition location because of concurrent memory requests.

• The observer process unexpectedly exits during the eviction of the plan cache because of the access to a null pointer.

• The observer process unexpectedly exits because of stack overflow, which is caused by a large number of static SQL statements in the anonymous block when prepared statement is disabled.

• The execution of the SELF ANTI JOIN statement results in an incorrect estimation of the number of rows because of SQL rewriting.

• The OB_ARCHIVE_TEMP_MEMORY module has memory leaks during log archiving.

• The PL statement cannot hit the plan cache.

• The OBServer retains the session after the client is disconnected.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP4

Version information

• Release date: July 13, 2021

• Version: V2.2.77 BP4

Enhanced features

• The MAKETIME() function is supported in MySQL mode.

• The UTC_DATE(), UTC_TIME(), and GET_FORMAT() functions are supported in MySQL mode.

Fixed issues

• XA transactions in the Active status cannot be rolled back.

• In Oracle mode, if the index column contains the CHAR data type and values that are less than or equal to 0, the index column cannot be queried.

• The execution of the ANTI-, SEMI-, and OUTER-JOIN statements with an OR join condition generates incorrect data.

• The observer process unexpectedly exits because of an OSS SDK bug during the physical backup to the Alibaba Cloud OSS bucket.

• The observer process unexpectedly exits sometimes because of queries on the gv$sysstat view.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP5

Version information

• Release date: July 31, 2021

• Version: V2.2.77 BP5

Enhanced features

• You can set the SQL_MODE parameter to NO_ZERO_IN_DATE in MySQL mode.

Compatibility changes

• The following default actions of the SQL_MODE parameter are changed in MySQL mode:

Fixed issues

• The observer process unexpectedly exits when an excessive amount of columns are operated by the TRIGGER operation.

• The query on the generated column in conjunction with the ORDER BY LIMIT action generates an incorrect plan.

• Incorrect results are generated because no deep copy is performed during a BATCH NESTED LOOP JOIN operation.

• The observer process unexpectedly exits because no deep copy that allocates the memory is performed when the PL engine executes dynamic SQL statements.

• The observer process unexpectedly exits because like Number is used as a filter condition.

• The observer process unexpectedly exits when the redo log module prints logs.

• An error is returned when the SHOW PROCEDURE STATUS statement is executed after the execution of the DROP USER statement.

• The plan_type field of the sql_audit parameter displays incorrect plan type.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP6

Version information

• Release date: August 25, 2021

• Version: V2.2.77 BP6

Fixed issues

• The backup set is marked as Deleted after the secondary backup fails.

• In a scenario where a tenant recovery operation is performed, if an existing tenant has the same name as the tenant to be recovered, the existing tenant is dropped.

• In a parallel execution (PX), if a long timeout is set for a tenant by default, the operation of session termination takes a long time to respond.

• The observer process may fail to start with a small probability after clog compression is enabled.

• Deadlocks are possible when SECURITY AUDIT is enabled in high concurrency scenarios.

• The observer process unexpectedly exits in Oracle mode because of the execution of the SELECT FETCH UNION SELECT FETCH syntax.

• The observer process unexpectedly exits when the memory request fails during the execution of a distributed plan.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP7

Version information

• Release date: September 18, 2021

• Version: V2.2.77 BP7

Enhanced features

• The eviction mechanism of SQL execution plans for partitioned tables is optimized.

• Secure Sockets Layer (SSL) communication encryption is supported by ob_admin.

• The LOAD DATA REPLACE syntax is supported in Oracle mode.

• The user logon and logoff statistics are collected.

• The ORDER BY expression is supported in MySQL mode.

• Automatic load balancing for partitions in a STOP SERVER/ZONE operation is optimized.

• The routing of temporary tables is optimized.

Fixed issues

• The creation of a tenant fails if the _enanble_easy_keepalive parameter is set to true.

• The substr(c1,x,1)='a' query cannot hit the plan cache.

• The logical recovery fails because the rowkey does not meet the length requirements.

• The accumulation of clogs causes the OBServer to restart, which results in excessive memory consumption by the sys tenant.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP8

Version information

• Release date: October 15, 2021

• Version: V2.2.77 BP8

Enhanced features

• The internal __ALL_VIRTUAL_PROCESSLIST table is provided to display the transaction timeout status.

Fixed issues

• Under the SQL parallel framework, the SELECT FOR UPDATE statement causes the optimizer to return the OB_STATE_NOT_MATCH error.

• The gv$SQL_PLAN_MONITOR view cannot properly display non-distributed plans.

• An internal error is returned when the NVL() function is used in the CASE WHEN statement.

• In Oracle mode, the value of the PARTITION KEY COUNT field in the ALL_PART_TABLES table is not displayed.

• The tenant recovery operation fails when the tenant is deleted.

• In MySQL mode, the execution of the SET PASSWORD statement in stored procedures does not take effect.

• The creation of views in PL fails in MySQL mode.

• The redo_transport_option parameter of the standby cluster does not take effect.

• The observer process unexpectedly exits during the conversion from the INT and STRING data types to the DATATIME, DATE, and TIME data types.

• The observer process unexpectedly exits during the process of setting the PARALLEL_MAX_SERVERS variable.

• In Oracle mode, the use of TRIGGER causes the SQL EXECUTOR module to have memory leaks.

• The 4019 error is returned when you create a tenant after a large number of tenants are deleted.

• The observer process unexpectedly exits during the process of setting the clog_disk_usage_limit_percentage and clog_disk_utilization_threshold parameters.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP9

Version information

• Release date: November 19, 2021

• Version: V2.2.77 BP9

Fixed issues

• The 4038 error is returned at the close phase of parallel execution.

• When PL debugging is performed in the OceanBase Developer Center (ODC) console, resources for debugging cannot be quickly released if ODC directly exits.

• The client session is disconnected when the memory request fails at the SQL PARSE phase.

• When a temporary table is created and added with constraints, the cluster is hung if a constraint of the temporary table has the same name as that of a non-temporary table.

• The MERGE INTO (SELECT *) statement is incompatible with Oracle.

• In Oracle mode, the ALL_TAB_PARTITIONS view displays partitions in an unordered way.

• The 4016 error is returned during the execution of the REPLACE INTO statement.

• The gv$sql_audit view displays inaccurate QUEUE_TIME and RETRY_CNT statistics.

• The incorrect FILE HANDLE is obtained when the NFSv4 protocol is used for backup.

• The gv$sql_audit view displays the password for user creation in plaintext.

• If the base table of the gv$sql_audit view does not exist, the value of elapsed_time in the recorded query information is 0.

• In Oracle mode, the pivot and unpivot clauses are incompatible with Oracle.

• The restart of the observer process keeps failing because it cannot be replayed after the restart, which is caused by the reuse of old slogs due to insufficient storage space.

• When the SHOW TABLE STATUS statement is executed to query a partitioned table, the result is NULL.

• In Oracle mode, the error unknown timezone is returned for the obclient connection after you specify the global time_zone system variable.

• In MySQL mode, the user privileges on a table remain valid after the table is deleted.

• Some IP addresses in the gv$sql_audit view are parsed into 0.0.0.0.

• In Oracle mode, an error is returned when the SELECT * FROM (GENERATED TABLE) statement is executed due to duplicate column names.

• The BIT data type is incompatible with MySQL.

• The observer process unexpectedly exits when it reads the configuration file during startup.

• The observer process unexpectedly exits during the data backup to OSS.

• The observer process unexpectedly exits during the conversion of the SET/ENUM/LOB data types.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP10

Version information

• Release date: January 17, 2022

• Version: V2.2.77 BP10

Enhanced features

• The tag cleanup mode is supported in OSS.

• The USERENV('SID') and USERENV('LANGUAGE') functions are supported in Oracle mode.

• When the backup space reaches 100%, you can delete obsolete backup files.

• The stability and performance of the queuing table are improved.

• The streaming compression algorithm, which is not supported for clog compression, is removed.

• The 4 KB aligned writes of disk DIRECT_IO are supported.

Compatibility changes

• By default, the auto_refresh_location_cache parameter is set to true to enable the balance strategy for partition replicas.

• By default, the schema_history_recycle_interval parameter is set to true to enable the periodic cleanup of expiry records in the system table by background threads.

Fixed issues

• The column type in the view of null columns is NULL, but that in the Oracle database is char(0).

• In the gv$sql_audit view, the maximum length of the QUERY_SQL column is not 64 KB.

• In Oracle mode, a false positive error is returned, which indicates that the total row length of all columns exceeds the value of the OB_MAX_USER_ROW_LENGTH parameter.

• The index creation fails when the GBK character set is used.

• The statistics of a tenant recovered by using the backup and restore feature are not promptly updated.

• The error not support is returned when you set the secure_file_priv parameter after the connection is created by using the IP address 127.0.0.1.

• An error is returned for incorrect calculation of string length when the CREATE TABLE statement involves the setting of variable values.

• The error Column length too big for column is returned during the length check of the LOB columns.

• An error is returned during the execution of the ALTER TABLE DROP COLUMN statement due to the delayed deletion of physical backups.

• When the sql_auto_is_null parameter is set to ON, the error Invalid Argument is returned during a query where the auto-increment columns need to be identified.

• In Oracle mode, no privilege is provided to control the TRUNCATE DLL operation.

• The creation of a global index fails if the process of an OBServer node exits.

• The __all_virtual_memory_info table displays incomplete tenant information.

• The try_cnt column of the gv$sql_audit view shows inaccurate statistics.

• In Oracle mode, the TO_CHAR() function cannot recognize the date format.

• An error occurs when checking is conducted for the index created on a column of the CHAR type in Oracle mode.

• All rows are NULL after the LEFT JOIN UPDATE on multiple tables.

• In PL mode, the observer process unexpectedly exits due to the existence of the PRINT LOB data type.

• The observer process unexpectedly exits because the LIKE clause in the SQL statement exceeds 250 KB in length.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP11

Version information

• Release date: May 2, 2022

• Version: V2.2.77 BP11

Enhanced features

• The integration of the transparent data encryption (TDE) feature with the Alibaba Cloud Key Management Service (KMS) system is supported.

• The performance of non-blocking Table API queries is improved.

• The maximum value of the in(values) query is optimized to make it compatible with MySQL.

• Filtering by tenant is supported by liboblog to synchronize only logs of the specified tenants.

• Returning the synchronization timestamps of logs is supported by liboblog.

Fixed issues

• In MySQL mode, an error is returned during the parsing of the GROUP BY expression because of unclear semantics. In this version, the expression from the selection list is used to avoid errors.

• In MySQL mode, an error is returned for invalid characters generated during the conversion of the GBK character set by calling the CONVERT() function. In this version, the invalid characters are converted to 0x3F, which is compatible with MySQL.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.

• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP12

Version information

• Release date: June 14, 2022

• Version: V2.2.77 BP12

Fixed issues

• In Oracle mode, when you create a stored procedure with the specified character set GBK, the OBServer becomes suspended unexpectedly during the SQL prepare phase.
• In MySQL mode, an error is returned because the Limit operator does not support character type conversion.
• When the system performs a follower-to-leader switchover during a major compaction, the SQL statement is rolled back because the cached route is inaccurate. This causes a long execution time.
• The error code 4007 is returned when the DBMS_LOB.SUBSTR system package is executed in Oracle mode.
• The queue_time field is displayed as a negative value in the v$sql_audit view.
• In the sys tenant, the resource specifications are not restored to the default settings after a unit is removed.
• When the input parameter is converted during the execution of a stored procedure, the following error is returned: value too large for column.
• In multi-table join queries, the CPU utilization is high when the join key has an index and a ROWNUM or LIMIT operator.
• When a null condition is used to query the INFORMATION_SCHEMA.COLUMNS view, incorrect results are returned.
• The observer process unexpectedly exits due to an exception occurring in table access by global index primary key. This is caused because the background thread does not initialize the tenant context.
• The observer process unexpectedly exits because the cluster is stopped while data is written to Tencent Cloud Object Storage (COS) or Alibaba Cloud Object Storage Service (OSS).
• A syntax error is returned when the DBMS_LOB system package is executed to insert data.
• The logs of the Root Service node repeatedly report errors because an exception occurs when an index is created.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.
• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP13

Version information

• Release date: August 26, 2022

• Version: V2.2.77 BP13

Enhanced features

• Adding partitions is supported in specific scenarios. You can add partitions at the beginning or in the middle of RANGE-partitioned tables. When you add a partition, make sure that no data is written to this partition. You can enable this feature by using the tenant-level parameter _enable_add_between_range_partitions.
• Cluster-level parameters of a specific OBServer can be modified by using ob_admin. However, tenant-level parameters cannot be modified for a specific OBServer.
• Backup files on Tencent Cloud COS can be tagged to facilitate file management. Fixed issues
• The cluster restart times out due to an excessive number of schema history records.
• In MySQL mode, an error is returned for invalid characters generated during the conversion of the GBK character set by calling the CONVERT() function. In this version, the invalid characters are converted to 0x3F, which is compatible with MySQL.
• In the primary-standby deployment mode, records in the virtual table __all_server_event_history expand fast due to frequent execution of the ALTER SYSTEM CHANGE TENANT operation on the primary database.
• During the execution of a distributed plan, when one OBServer fails, the execution of SQL statements on other OBServers hangs until timeout.
• When an equivalent condition is used to detect the case sensitivity of the collation, the returned result set is wrong because an execution plan is incorrectly hit.
• When a backup task is forcibly canceled, the backup status stays in Stopping.
• When the INSERT...ON DUPLICATE KEY UPDATE statement is executed, an error indicating duplicated primary keys is returned.
• An error is returned during the execution of a nested aggregate operator that contains Distinct.
• Invalid log information about RPC communication between OBServers is deleted.
• When the __min_full_resource_pool_memory parameter is set to an abnormal value, the OBServer can still be started.
• When the LAST_INSERT_ID() function is called within a stored procedure, a invalid execution result is generated.
• The error code 4016 is returned when a partition is truncated.
• An error occurs for the data of the net_wait_time field in the gv$sql_audit view.
• When the weak_read_version_refresh_interval parameter is set to 0, the data in partitions is not synchronized.
• The query timeout hint added to the cursor of a stored procedure does not take effect.
• In Oracle mode, when a prepared statement is enabled, the return value of the NUMTODSINTERVAL() function is truncated to a specific precision.
• After a partition is dropped or truncated, the locality change is suspended due to duplicate global indexes.

Upgrade and limits

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.
• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.

V2.2.77 BP14

Version information

• Release date: December 04, 2022
• Version: V2.2.77 BP14

Enhanced features

• You can change the AccessKey ID and AccessKey Secret during data backup and restore. The backup feature of OceanBase Database allows you to specify an Alibaba Cloud Object Storage Service (OSS), Tencent Cloud Object Storage (COS), or Huawei Object Storage Service (OBS) bucket as the backup destination. However, you cannot modify the access_id or access_key parameter of the backup destination during the backup. If the AccessKey ID or AccessKey Secret is changed, you must stop the log backup and change the backup directory. Therefore, this version provides the ALTER SYSTEM change external_storage_dest PATH[=]'xxx?host=xxx' set ACCESS_INFO[=]'access_id=xxx&access_key=xxx'; command for you to modify the AccessKey ID and AccessKey Secret of the backup destination. The relevant content in GConf is updated and saved accordingly. Observe the following considerations when you use this command:

  • Before you run the command to update the AccessKey ID and AccessKey Secret for accessing the OSS, COS, or OBS bucket, we recommend that you stop the ongoing second backup task. We also recommend that you delete the old AccessKey ID and AccessKey Secret one hour later after you run the command.
  • If you modify the AccessKey ID and AccessKey Secret during a second backup task, and the task is stuck, you must forcibly cancel the task and change the directory to resume the task.

• The following three fields are added to the v$sysstat view to optimize hotspot row observation.

  • TRANS_ELR_ENABLE_COUNT: the number of transactions for which early lock release (ELR) can be enabled.
  • TRANS_ELR_UNABLE_COUNT: the number of transactions for which ELR is disabled.
  • READ_ELR_ROW_COUNT: the number of rows that support ELR. The number of rows that support ELR is increased.

• ob_admin supports OBS.

Fixed issues

• The execution of the to_hex() function during stored procedure debugging causes server failures.
• When system variables fail to be loaded, the session is abnormally released, and this causes a server failure.
• The time zone used by the backup logic is different from that used by the restore logic. As a result, the default values of restored table fields lag behind those of the original table by 8 hours.
• When the _enable_fulltext_index parameter is set to OFF, creating a full-text index succeeds, which is abnormal.
• When the special characters "+/" are used in a LIKE statement, the query result of an index scan is incorrect.
• When an inner join subquery is executed, the query results in OceanBase Database and MySQL Database are inconsistent.
• In an outer join, the number of rewritten rows is inconsistent with the number of queried rows.
• The number of constraints on the CONCAT() function is negative, which is abnormal.
• When the batch deletion operation of COS is called, the system cannot determine whether all objects are deleted.
• When you delete a large number of Fail to erase key from cache entries from the observer.log file, an error is returned.
• In non-strict mode, the execution result of creating an index is different from that of adding an index.
• The use_hash_aggregation hint does not take effect.
• After schema history file recycling is enabled, physical restore may fail.
• The backup may fail due to an incorrect session count in the independent backup and restore subprocess.
• During the procedural language (PL) execution by using the OCI_BATCH_ERRORS handle, a data storage error occurs due to primary key conflicts.
• If slogs and SSTables are located on the same disk, the clog disk is almost full error is returned during slog initialization.

Upgrade and limitations

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.
• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.
• If you have enabled the dynamic AccessKey ID/AccessKey Secret setting feature for data backup and restore, an upgrade to V3.x is not supported.

V2.2.77 BP15

Version information

• Release date: January 10, 2023
• Version: V2.2.77 BP15

Enhanced features

The FROM_BASE64() and TO_BASE64() functions are supported in MySQL mode.

Compatibility changes

In MySQL mode, the ORDER BY or LIMIT operator cannot be used for multi-table updates, which is consistent with MySQL.

Fixed issues

• When you use an IN expression to query a partitioned table, the calculated result set is incorrect.
• The migration of an empty resource unit causes the leader to get stuck.
• When an execution plan hits the plan cache, the execution result of MERGE INTO is incorrect.
• Memory leak occurs when indexes are created on JSON generated columns.
• Backup sets fail to be cleared during automatic backup clearing.
• In Oracle mode, when a package uses a cursor for loop operations, the error code 4179 is returned.
• The standby database fails to restart due to an slog replay error.
• Errors are returned when the LPAD function processes special characters whose ASCII codes are 0x03, 0x04, 0x05, and 0x06.
• During the clearing of expired backups, the clearing task remains in the DOING state and cannot be stopped.
• If the execution of a transaction is interrupted, the transaction ID recorded in SQL audit logs is incorrect.

Upgrade and limitations

• The primary and standby clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must switch the cluster to maximum performance mode before you can upgrade the cluster.
• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.
• If you have enabled the dynamic AccessKey ID/AccessKey Secret setting feature for data backup and restore, an upgrade to V3.x is not supported.

V2.2.77 BP16

Version information

• Release date: February 14, 2023
• Version: V2.2.77 BP16

Enhanced features

The GROUP COMMIT capability is supported for OBKV to update hotspot rows. To better handle highly concurrent updates of hotspot rows, the server that supports GROUP COMMIT aggregates concurrent updates to prevent repeated row lock application and release for each operation. This saves the wait time due to row lock contention, and speeds up the hotspot row update.

Fixed issues

• The execution of the USE DATABASE statement triggers implicit commit of transactions.
• The SELECT statement in PL cannot read the data that has not been committed.
• Some indexes are abnormal after a successful physical restore.
• When you create a view, you must also specify an alias for ROWID in the SELECT clause even if you have specified column names.
• The BATCH GET request of HBaseAPI takes a long time to complete.
• The update of a primary key column in a TableAPI operation results in data inconsistency between the primary table and the index table.
• An error is reported when TableAPI automatically updates fields with default values of CURRENT_TIMESTAMP or ON UPDATE CURRENT_TIMESTAMP.
• An error is reported when the estimated result length of the REGEXP_REPLACE function is greater than the maximum allowable value because of excessively strict program check.
• The binary comparison of the FIND_IN_SET(binary 'b','a,B,c,d') function does not take effect in MySQL mode.
• The execution of the FIND_IN_SET function is slow in MySQL mode.
• The backup_piece files that are expected to be cleaned up are not deleted in a normally scheduled automatic cleanup task.

Upgrade and limitations

• The primary and secondary database clusters can be upgraded only in maximum performance mode. If a cluster is in maximum protection mode or maximum availability mode, you must downgrade the cluster to maximum performance mode before you can upgrade the cluster.
• If you have enabled the dynamic AccessKey ID (AK)/AccessKey secret (SK) setting feature for data backup and restore, you cannot upgrade OceanBase Database to V3.x.
• PL debugging is supported only for x86 platforms. We recommend that you do not use this feature on an ARM platform.