The WarehousePG Enterprise Manager (WEM) 1.3.x documentation describes the latest patch release. Release notes for all 1.3.x versions are included below.
| Version | Release date |
|---|---|
| 1.3.0 | 10 Aug 2026 |
WEM 1.3.0
Released: 10 Aug 2026
WEM 1.3.0 includes new features, enhancements, and bug fixes.
Highlights
Host agent: WEM 1.3 provides a new
acp-host-agentpackage, a separate service that runs on the WarehousePG coordinator host and executes privileged operations on behalf of WEM. The host agent connects to WEM's internal job-execution engine, which makes operations reliable and recoverable even when connectivity is interrupted. See Installing the host agent.Bundled Alertmanager: WEM now runs Alertmanager as an internal managed subprocess rather than requiring a separately installed and configured instance. See Configuring WEM.
Fully agent-based HBA editor: The HBA editor now operates through the host agent regardless of whether WEM is installed on the same host as the WarehousePG coordinator. See Modifying authentication rules.
New features
- System Metrics improvements: The System Metrics views now support filtering by individual node, so you can isolate CPU, memory, and disk usage for a specific host. Time-range controls are also refactored, and charts adapt their formatting to the selected time span. See Visualizing hardware performance.
- Storage date range picker: The Storage > Historical tab now includes a date range picker for precise control over the time window for storage trend analysis. See Planning storage capacity.
- Data Analysis improvements: WEM now uses dedicated database views to compute object sizes, replacing per-object size queries and reducing overhead when browsing large schemas. On WHPG 7 clusters, Admins can also switch skew calculations from a fast catalog-based estimate to an accurate mode that scans the file system for exact figures. See Analyzing data distribution.
- Query Monitor improvements: A new chart on the Longest Queries tab shows the top 50 longest-running queries over time, making it easier to spot and investigate persistent slow queries. A new Historical Queries tab shows completed queries along with a cumulative query count. See Identifying the longest-running queries and Monitoring and evaluating queries.
- Log viewer improvements: Log views now display the User and Source Host for each log entry, making it easier to trace activity back to a specific database user or client machine. The log viewer also supports fetching all matching records through server-side pagination, with a live progress counter and the option to cancel a long-running fetch, and exported CSV files match exactly what's shown in the grid. See Auditing system logs.
- Alert rule configuration: Previously, creating or editing an alert rule meant editing a Prometheus rule YAML file directly. In 1.3, the Rules tab in the Alerts panel lets you create and edit rules through a builder (metric, comparison operator, threshold, and label filters), or as raw PromQL. Built-in system rules can't be deleted by anyone, including Admins. Admins can edit or delete any user-created rule, but Operators can only manage their own. The Notifications tab now lets you edit the Alertmanager configuration file directly to set up Slack, email, or other receivers. The History tab adds a rule-name filter and an Admin-only option to delete history, either entirely or by date range. See Managing alerts.
- Password reset links and CLI command: Admins can generate single-use password reset links for users from the Management panel. A new
wem reset-passwordCLI command supports local or offline password resets and automatically loadswem.confin standard installations. See Performing administrative management.
- pg_hba.conf tab improvements: The pg_hba.conf tab in Access Management distinguishes cluster-internal trust rules from broad-network trust rules, making it easier to spot overly permissive entries. Saving changes also now propagates the update to the standby coordinator automatically, keeping both coordinators consistent. See Modifying authentication rules.
Changes
- Canary Checks panel: The Monitoring panel is renamed to Canary Checks. Configuration of these checks, previously Admin-only in the Management panel, is now consolidated here. Operators and Viewers can view results. Admins can create, edit, and delete checks. See Validating database responsiveness.
- Alertmanager configuration: The
ALERTMANAGER_URLandCONFIG_PATHconfiguration parameters are removed, since Alertmanager is now bundled and requires no external URL. If you're upgrading, remove these from your existingwem.conf. See Upgrading WarehousePG Enterprise Manager.
- Longest Queries performance: WEM now caches longest-running query data from the external table on a background schedule instead of scanning it on every request, cutting typical lookup times from tens of seconds to milliseconds. See Identifying the longest-running queries.
Bug fixes
- Fixed the canary trend chart showing an incorrect Y-axis scale. The chart now correctly shows a weighted success ratio.
- Fixed pagination behavior across UI tables: pagination now resets correctly when filters change, and page controls are consistent across all views.
- Fixed the dashboard failing silently when
WEM_COOKIE_SECRETis unset. WEM now validates the cookie secret at startup and surfaces a clear error.
- Fixed host agent installation to handle local authentication scenarios for the configuration database connection, and to work with new installation paths.
- Fixed the pg_hba.conf tab to match the backend's trust-rule classification, and to refresh HBA statistics immediately after saving.
- Fixed issues editing and saving the Alertmanager configuration file. Changes that fail to reload now roll back automatically, so the saved file always matches the running instance.
Fixed table size and compression estimates in Data Analysis to compute from catalog metadata, improving accuracy and performance on large schemas.
Fixed alert history and the users audit log to no longer cap results at 500 rows.
Fixed the partition details view to no longer cap results at 500 rows, to show the immediate parent table instead of the root table for multi-level partitions, and to correctly count partitions and indexes across all levels.
Fixed timeouts on Data Analysis skew and top-tables queries on large clusters.
Fixed the database size collector to run in the background instead of blocking the request that triggers it.
Fixed user creation to correctly apply the Active status from the request.
Fixed a race condition where saving
pg_hba.confto the standby coordinator returned success before the operation finished.