Skip to main content

DBX Agent Release Notes

2026-10-22

v2.64.05

MS SQL Server, MySQL, PostgreSQL, Oracle

Feature Added per-database collection of the lock wait session count (lock_wait_sessions).

  • Applies to instances with multiple databases.
  • The existing instance-wide value is still collected.
  • Databases without lock waits send 0.
Oracle

Fixed Fixed lock information not being collected on Oracle 10g.

DB2

Feature Added buffer pool, log, sort, catalog cache, and wait time metrics.

  • Target: db_db2_counter, db_db2_bufferpool
  • Buffer pool: bp_cur_buffsz (current size per buffer pool), no_victim_buffer
  • Log (DB2 10.1 or later): total_log_size, sec_logs_allocated, log_read_time, log_write_time, log_to_redo_for_recovery, log_held_by_dirty_pages
  • Sort, hash join, OLAP (DB2 10.5 or later): total_sorts, sort_heap_allocated, sort_shrheap_top, active_hash_joins, active_olap_funcs
  • Catalog cache (DB2 10.5 or later): cat_cache_lookups, cat_cache_inserts
  • Wait time (DB2 10.5 or later): total_wait_time

Fixed Fixed the log read/write metrics (log_reads, log_writes, log_time) always being collected as 0.

Fixed Fixed active_sorts being collected as an increment instead of the current value.

Changed Changed cat_cache_ratio from the catalog cache miss ratio to the hit ratio (%).

  • Intervals with no lookups send 100.

2026-10-18

v2.64.04

SingleStore

Changed Improved the replication topology collection structure.

  • Topology collection now runs in a dedicated thread, so 5-second metrics (active sessions, counters, etc.) are not delayed even when topology queries take long on large clusters. This adds one DB connection session.
    • topology_interval: Topology collection interval in seconds (default: 5).
    • topology_heartbeat_interval: Interval in seconds for sending a full snapshot even when nothing changes (default: 300).
    • topology_dbcount_interval: Interval in seconds for refreshing partition and table counts per database (default: 60).
  • The collection interval adjusts automatically to the collection time (up to 300 seconds) and returns to normal on recovery.
  • Topology queries are split to reduce DB load. The sent data (rows, fields, tags) is unchanged.
  • Sending is skipped when the data matches the previous collection. While replication lag values (lsn_lag, volume_lag_mb, catchup_s) are changing, data is sent on every collection.

Feature Added an option to stop pipeline collection after consecutive failures.

  • When pipeline statistics or error collection fails the specified number of times in a row, that collection stops. The failure count resets on success.
  • pipeline_error_limit: Number of consecutive failures allowed (default: 5). Set to 0 or less for no limit.

Feature Added a minimum table size option for columnstore compression ratio calculation.

  • compression_min_bytes: Lower limit of the original size, in bytes, for compression ratio calculation (default: 1048576 (1 MB)). Set to 0 for no lower limit.
  • Tables smaller than the lower limit do not send the compression ratio (compression_percent).
  • Prevents fixed segment overhead from distorting the compression ratio of small tables.

Changed Removed the os_other_used_mb and os_other_pct metrics from the node counter (db_singlestore_counter).

Fixed Fixed monitoring collection queries appearing in the active session list.

2026-10-06

v2.64.03

MySQL

Fixed Fixed an issue where timer_wait and lock_time values in SQL statistics were collected 1,000 times smaller than the value in seconds.

MySQL, PostgreSQL

Fixed Fixed an issue where SQL statistics were collected starting from low-load SQL.

  • SQL statements with the longest execution time are now collected first, up to statements_row_limit rows.

Fixed Fixed an issue where negative values were collected in SQL statistics after a DB restart or statistics reset.

PostgreSQL

Fixed Fixed an issue where SQL statistics were not collected on PostgreSQL 17.0.

Changed If no sort direction is specified in statements_order_by, results are now sorted in descending order (desc).

Oracle

Feature Added an option to disable transaction information queries during active session collection.

  • Option: ora_txn_undo (default: true)
  • When set to false, the agent does not query v$transaction and does not collect undo information (undo_segid, undo_blk, undo_rec).
  • Restart the agent to apply the change.
Common

Fixed Fixed an issue where the Windows service installation script (install_WindowsService.bat) stopped in paths containing parentheses (e.g., C:\Program Files (x86)).

Fixed Fixed an issue where the agent registered as a Windows service did not start.

  • When JAVA_HOME (or JAVA_BIN) was enclosed in quotation marks (e.g., "C:\Program Files\Java\jdk"), service registration succeeded but the agent stopped immediately after starting.
  • Also fixed an issue where the Java check step of the installation script accepted invalid paths.

2026-10-02

v2.64.02

CUBRID

Fixed Fixed an issue where the tablespace size metric was collected as a negative or invalid value when it exceeded 2 GB.

2026-09-30

v2.64.01

DB2

New Added real-time CachedPlan lookup.

  • View plans without running Explain Plan.

New Added CachedPlan saving.

  • db2_plan_saving: Whether to save plans (default: true)
  • db2_plan_elapse: Minimum execution time of SQL to save, in seconds (default: 5)
  • db2_plan_max_count: Maximum number of lookups per collection interval (default: 50)

New Added Plan Change collection.

Feature SQL statistics are now separated by execution plan.

SingleStore

Feature Expanded replication topology collection data and added a collection interval option.

  • Partition replication and reference replication are collected together.
  • topology_interval: Collection interval in seconds (default: 5)

Feature Added support for SingleStore 7.3.

Feature Changed table statistics collection from once a day at a specified time to interval-based collection.

  • table_interval: Collection interval in hours (default: 1)
  • The tables_hour option is no longer used.

Feature Unified the collection interval unit for per-partition table statistics to hours.

  • row_table_interval: Default 1 (previously 5 minutes)
  • partition_interval: Default 12
  • column_table_interval: Default 1

2026-09-02

v2.63.06

Oracle

Fixed Fixed an issue where statistics for subpartitioned indexes were not collected.

MySQL

Fixed Fixed an issue where SQL statistics (statements) were not collected in long-running environments.

  • Collection stopped when the cumulative SQL execution time in performance_schema exceeded a certain value.
  • Occurred on databases that had been running for a long time without a restart.

2026-08-26

v2.63.05

Common

Feature Added support for viewing and editing agent configuration from the screen.

  • Check whatap.conf options and change their values on the agent configuration screen.
  • Comments, option order, and the existing notation of the configuration file are preserved on save.
  • Deleting an option removes it from the configuration file and the agent falls back to the default value.

Feature Added support for restarting the agent from the screen.

  • Apply configuration changes by restarting from the screen without connecting to the server.

Fixed Fixed an issue where values containing a backslash were saved incorrectly.

  • For example, C:\data\whatap was applied as C:datawhatap.
  • Applies to options that take a path as their value (dbx_log_path, cms_token_file, fileview).

2026-07-29

v2.63.04

SingleStore

Changed Node state is now sent separately under the dedicated db_singlestore_node_state category.

  • Sent every 5 seconds per node, providing the node state values of the cluster view without further processing.
  • You can check the state of every node, including aggregator nodes that hold no partitions.
  • The node state metrics in the existing db_singlestore_counter (fatal_count, warning_count) have been removed.

2026-07-28

v2.63.03

Common

Changed Added a log entry showing which credential path (static key / AssumeRole / EC2 Role) was used to connect during IAM authentication.

PostgreSQL, MySQL, MariaDB

Feature Added support for connecting to RDS/Aurora with an AWS IAM token instead of a password.

  • Options: aws_iam_auth=true + aws_region
  • Credentials: static key / AssumeRole (aws_arn) / EC2 instance role
MongoDB

Feature Added support for IAM authentication (MONGODB-AWS) on Amazon DocumentDB 5.0+.

  • Upgraded the MongoDB driver to 4.11.5 (IAM authentication requires 4.1+)
  • Connects with the IAM user/role registered in $external
Redis

Feature Added support for IAM authentication on Amazon ElastiCache (Redis 7+ / Valkey 7.2+).

  • Options: aws_iam_auth, elasticache_name
  • Supports both serverless and node-based deployments (serverless does not support CLIENT LIST, so active session collection is excluded)

2026-07-27

v2.63.02

Tibero

Fixed Fixed an issue where the SQL child_number value was collected incorrectly when it exceeded the int range (now handled as a long).

SingleStore

Feature Added per-node state monitoring for the cluster.

  • Collects the state of each node (MA / CA / LEAF) as online / attaching / down.
  • A down node is reported as fatal_count=1 and an attaching node as warning_count=1.
  • Data is sent separately for each node (routed by node OID, including heartbeat).

Fixed Fixed an issue where MySQL SQL statistics were not collected.

  • Affected versions: 2.59.01 (2026-01-21) ~ 2.61.02

2026-07-22

v2.63.01

MySQL, PostgreSQL

Feature Added database replication topology collection.

  • Collects the replication structure (cluster grouping), node state, and cause information to visualize the replication setup on the topology screen.

Feature Added automatic detection of the cluster grouping key.

  • When the existing replication_name option is not set, the grouping key is detected from the replication setup and the cluster is grouped.
  • Auto-detection supported: MySQL 5.6+ (based on server_uuid), PostgreSQL 9.6+ (based on system_identifier), Galera, Group Replication.
  • Auto-detection not supported (option must be set as before): Aurora MySQL/PostgreSQL, RDS/Azure PostgreSQL, MariaDB traditional replication, MySQL earlier than 5.6, PostgreSQL earlier than 9.6.
  • When the option is set, its value takes precedence as before (auto-detection is disabled; set the same value on all cluster nodes).

Changed Instance list sets are now grouped automatically.

  • Instance sets are built from the automatically detected grouping key even when replication_name is not set (previously they were grouped only when the option was set).
MySQL

Changed Cluster grouping now follows failover automatically.

  • Detects master promotion/redirection and refreshes the grouping key automatically (converges within a few seconds to 1 minute).
  • A multi-source slave subscribing to several masters is grouped by its first channel.

Feature Added support for the Galera cluster topology.

  • Reports quorum loss (non-Primary), cluster departure, and query rejection as failures, and collects cluster state metrics.
PostgreSQL

Feature Collects replication details per standby from the primary.

  • Collects replication state, synchronization mode, and delay information for each standby, matched automatically by replication slot (based on db_ip when no slot is used).

Fixed Fixed an issue where replication_delay was shown as an abnormally large value while replication was broken.

  • No value is sent during intervals where delay cannot be measured (right after a standby starts, delayed replication, and so on).

Fixed Fixed an issue where a broken standby replication state was sent as normal.

Fixed Fixed an issue on Aurora PostgreSQL where replication state was never collected if instance identification failed when the agent started (recovers automatically on reconnection).

2026-06-29

v2.62.03

SingleStore

Feature Bundled the JDBC driver with the agent (no separate local jar file required).

2026-06-24

v2.62.02

MS SQL Server

Fixed Fixed an issue where topology data was not collected when AlwaysOn was active on SQL Server 2012 and later.

DB2

Fixed Fixed an issue where wait class metrics were not collected.

2026-06-23

v2.62.01

Common

Feature Added support for AWS IAM token-based database authentication.

  • Connects using an AWS IAM authentication token instead of a database password (for MySQL / MariaDB / PostgreSQL RDS and Aurora).
  • Option: aws_iam_auth, default: false.
  • When set to true, connects using an IAM token; when false (default), keeps the existing password-based method.
  • When branching on aws_arn, the access/secret key can be entered directly in whatap.conf.
SingleStore

Feature Added new support for monitoring SingleStore databases.

  • Recognizes the distributed node structure (Master Aggregator / Child Aggregator / Leaf) and collects metrics per node.
  • Collects active sessions, system information, parameters, pipelines, group count status, tables/partitions, and more.
  • Supports control commands: force-kill a session, view the execution plan.

v2.61.04

Common

Fixed Fixed an issue where the agent script did not run on some Unix environments, including AIX.

Fixed Fixed an issue where the agent did not start on AIX (IBM J9 JVM).

2026-06-18

v2.61.03

MySQL

Fixed Fixed an issue where MySQL SQL statistics were not collected.

  • Affected versions: v2.59.01 (2026-01-21) to v2.61.02.

2026-06-01

v2.61.02

MySQL

Feature Added MariaDB Connector/J driver support for environments using MariaDB-native authentication (e.g. ed25519).

  • Option: mariadb_driver, default: false.
  • When set to true, connects using the dedicated MariaDB driver (org.mariadb.jdbc), enabling monitoring of ed25519-authenticated accounts that the existing MySQL driver could not support.
  • Requires placing the MariaDB Connector/J (3.x) driver file under jdbc/ in the install path.
  • For environments that require SSL (e.g. require_secure_transport=ON), use together with connect_option=?sslMode=TRUST.

2026-05-27

v2.61.01

DB2

New Added Wait class (wait event) metrics collection

New Added Parallel Query (PQ) tree collection

  • Visualizes subagent and coordinator structure for parallel SQL execution.
  • Additional monitoring account permissions required:
    • MON_GET_WORKLOAD
    • MON_GET_PKG_CACHE_STMT
    • WLM_GET_SERVICE_CLASS_AGENTS
    • MON_GET_DATABASE_DETAILS
    • MON_FORMAT_XML_WAIT_TIMES_BY_ROW
    • MON_GET_UNIT_OF_WORK
MS SQL Server

New Added Always On AG (Availability Group) topology collection

  • Visualizes Primary/Secondary replica status and per-AG sync state.

New Added automatic Failover detection

  • Immediately sends new state via db_set_info when replica role changes.

2026-05-22

v2.60.12

Common

Changed Added automatic service restart after agent auto-update in Windows Service.

  • Previously, the service remained stopped after auto-update, requiring a manual restart.

Changed Migrated Windows Service registration to NSSM-based method.

  • The agent (Java) process tree is fully terminated when the service stops, resolving the issue of processes remaining after termination.
  • Resolved shutdown failures frequently occurring in environments with Oracle Java installed.
PostgreSQL

Fixed Fixed an issue where table_bloat_bytes and index_bloat_bytes settings were not applied.

  • Resolved an error where user-defined values were ignored due to an incorrect key reference.

2026-05-12

v2.60.11

Oracle

Feature Added Oracle NLS parameter collection (ora_nls_param).

  • Default: true
  • When set to true, sends v$nls_parameters results (NLS_LANGUAGE, NLS_TERRITORY, NLS_DATE_FORMAT, NLS_CHARACTERSET, NLS_SORT, etc.) along with existing parameter collection.

2026-05-08

v2.60.10

MySQL

Feature Added support for Aurora cascade topology — automatically collects slave status (db_mysql_slave) when an Aurora Writer receives binlog from an external master.

  • No additional configuration required. Cascade topology is detected automatically.

Fixed Fixed a false alarm when Seconds_Behind_Master is NULL.

  • Resolved a case where replication disconnect (NULL) was incorrectly displayed as normal.

2026-04-27

v2.60.09

CUBRID

Feature Added an option to collect Active Session SQL Text directly from the CUBRID Manager (gettransactioninfo) response (cub_disable_xcub_sqltext).

  • Default: false
    • false: Existing behavior — uses SQL Text delivered by the xcub agent.
    • true: Bypasses xcub SQL Text and uses the SQL_Text field from the CUBRID Manager response directly. Allows SQL Text collection without xcub on newer CUBRID versions.
  • Note: xcub must keep running even when set to true. To reduce xcub packet overhead, set tranlist=0 in xcub.conf.

Feature Added a debug log option for Active Session SQL Text collection (cub_sqltext_debug).

  • Default: false
  • When set to true, logs the raw SQL Text, sql_hash, and sql_param per session every collection cycle.

2026-04-17

v2.60.08

MS SQL Server

Feature Added an option to exclude specific accounts from active sessions.

  • Option name: skip_user
    • Default: empty (not applied)
    • When configured, excludes the specified accounts from both the active session list and session count.
    • Separate multiple accounts with a comma: skip_user=NT AUTHORITY\\SYSTEM,monitor_user
    • For Windows account names, enter backslashes as \\ (Java Properties escape rule).
Oracle

Fixed Fixed an issue where tablespace collected used > usagesize.

  • When a data file's size is set larger than maxsize (e.g., by manual RESIZE), the size value is used for usagesize.

2026-04-14

v2.60.07

Oracle

Fixed Fixed an issue where table size collection stopped after the first collection following agent startup.

  • Issue introduced in v2.60.02.

2026-04-13

v2.60.06

Oracle

Fixed Fixed underreporting of tablespace usage (used).

  • Fixed an issue where used was displayed as 0 for data files with no free space.

Changed Excluded OFFLINE data files from tablespace collection.

  • Prevents collection errors caused by inaccessible OFFLINE files.

Feature Added ASM Disk capacity collection.

2026-04-10

v2.60.05

Oracle

Feature Added the skip_whatap_session option.

  • When set to true, hides the whatap session from the active session list.

2026-03-26

v2.60.04

Tibero

Fixed Changed the data type of the pga_used_mem metric.

2026-03-23

v2.60.03

SQLMON

Feature Added the ability to specify a category per file.

2026-03-06

v2.60.02

Oracle

Feature Added the usagesize metric to the tablespace category.

  • Applies maxbytes if the data file is autoextensible (YES), or bytes if not (NO).
  • Calculates the tablespace usagesize by summing all values.

2026-03-05

v2.60.01

PostgreSQL

Feature Added an option to configure whether to collect table and index hit ratio.

  • Option: collect_hit_ratio_metrics
  • Default: true
DB2

Changed Changed the DB2 collection query to retrieve only the current DB instance information instead of all member information.

2026-02-11

v2.59.04

MongoDB, Redis, DB2, SAP ASE

Feature Added delta option

  • Option name: delta
  • Default: 2
  • Configures how delta values calculated per default collection cycle are stored.
    • 1: Converts the calculated delta value to a per-second rate and stores it
    • 2: Stores the delta value as-is based on the default collection cycle
SAP ASE

Fixed Fixed metrics: PagesRead, RowsDeleted, RowsUpdated

DB2

Fixed Improved hit ratio metric calculation method

  • Option name: includePrefetch
  • Default: true
  • false: Maintains the existing method
    • (logical_reads - physical_reads) / logical_reads * 100
  • true: Calculates based on synchronous page hits (pages_found_sync)
    • pages_found_sync / logical_reads * 100

2026-02-10

v2.59.03

Oracle

Fixed Fixed duplicate index information issue when collecting table layouts

PostgreSQL

Feature Added retry feature up to 3 times at 30-minute intervals when table size collection times out

2026-02-09

v2.59.02

MySQL, MS SQL Server, CUBRID, Altibase, Tibero

Feature Added delta option

  • Option name: delta
  • Default: 2
  • Configures how delta values calculated per default collection cycle are stored.
    • 1: Converts the calculated delta value to a per-second rate and stores it
    • 2: Stores the delta value as-is based on the default collection cycle

2026-01-21

v2.59.01

MySQL

Changed Improved MySQL statistics (db_mysql_statements) collection

  • Fixed overflow issue in existing collection criteria

Previous

sum_rows_affected+ sum_rows_sent+ sum_rows_examined>10000

Changed

sum_rows_affected>=10000
OR sum_rows_sent>=10000
OR sum_rows_examined>=10000
  • Added limit option to control the number of collections
    • statements_row_limit = 5000
  • Improved initial collection load by removing unnecessary sql_text queries during the first SQL statistics collection stage immediately after agent startup

Changed Fixed boundary value omission issue by changing limit condition from > to >= when collecting Table Size

PostgreSQL

Changed Improved PostgreSQL statistics (db_postgresql_statements) collection

  • Improved initial collection load by removing unnecessary sql_text queries during the first SQL statistics collection stage immediately after agent startup

Changed Improved frozenxid, seq_scan, idx_scan collection from instance-level collection to per-database collection method

Oracle

Changed Removed _no_5m_hour field to generate summary data for db_oracle_tablespace

Last updated on