v8
v8.3.0
Released on 18/11/2026
Overview
This release adds lookup tables: reference data you manage in Historian and read from calculations. The time series cache can now answer user and Grafana queries and keep the last value of every measurement, which takes load off the time series database. Event filtering and the per-event-type SQL views are much faster on large databases. The REST API validates requests more strictly, and several of those changes need attention before upgrading (see Before You Upgrade).
Lookup Tables
- Added lookup tables: reference data in a table with typed columns (String, Number or Boolean). Manage them on the new Lookup tables page in the main menu.
- Import rows from an xlsx or csv file, and export a table to an Excel workbook.
- Calculations and calculated event properties can read lookup tables and filter their rows.
- New privileges READ_LOOKUP_TABLES and MANAGE_LOOKUP_TABLES. Administrators and Managers get both, Readers get read access. Creating or changing tables needs the lookup tables license feature. Deleting a table does not.
- Memory limits for the tables held in memory: a warning limit (default 384 MiB) and a hard limit (default 512 MiB) in the general settings. Loading or writing a table that would cross the hard limit is refused.
- The table detail page shows which calculations and event properties reference it.
Events & Calculations
- Event configurations with a parent get a parent window setting: a child links when it starts inside the parent, or when its window overlaps the parent's. A wait for parent time sets how long a child waits for its parent before it is processed without one. Existing configurations behave as before.
- Detection modes have clearer names with a short description ("From timeseries", "From a schedule", "From users", ...).
- Filtering events on a property value and the per-event-type SQL views are much faster on large databases, and renaming a property keeps its values.
- Calculated event properties can be periodic (with or without dimension), with new script helpers for lookups and resampling. A script that returns the wrong type gets a clear error.
- A new calculation load view lists running calculations heaviest first, with load bars in the calculations page heading.
- The recalculate modal opens on Recalculate period, explains each option and marks Restart as advanced.
- The calculation editor warns about a sampled calculation without a time series input, and about an input that is an asset property without a measurement.
- Lower memory use on installations with many calculations.
Time Series Cache
- Added Serve user queries from the cache. With it on, time series queries from users and Grafana are answered from memory when the cache fully covers the range, including aggregations, group by on tags, descending order and open-ended ranges. Off by default.
- User query cache measurement limit caps how many measurements the cache keeps for user queries (empty means 1024).
- Added Cache last values. It keeps the newest value of every measurement in memory, so current-value panels and latest-value queries skip the time series database. The measurements table gets a Last value column, and the API can return last values on measurement and asset property lists. Off by default.
- The existing cache setting is now called Cache points. Cache points and Cache last values work independently. Serving user queries needs Cache points.
Assets, Prototypes & Measurements
- An asset property can exist without a measurement. Create one in the UI (the asset tree marks it with an unlink icon) or leave the measurement cell empty in an Excel import. Prototypes have a No measurement checkbox per property, and applying the prototype skips the event configurations and properties that would read it.
- A calculation that reads an asset property without a measurement is held, shows why on its detail page and on the homepage issues panel, and resumes once the property is linked again. Removing a property's measurement is refused while something other than a calculation reads it.
- Measurement names can be copied from the list and the detail panel.
API
- Stricter request validation. Write endpoints reject an empty body (an empty PUT could blank a label or reset an event). Names must be non-empty, UUIDs valid, limit and offset non-negative (a negative limit returned every row), and page at least 1.
- REST JSON request bodies are matched case-sensitively. A key in a different case (name instead of Name) is ignored like any unknown key. Set HTTP_JSON_CASE_SENSITIVE=false to restore the old behavior.
- REST request bodies are limited to 64 MB by default, configurable with HTTP_MAX_BODY_SIZE.
- The dependencies and dependents of a resource only show names of resources the user may read.
Users, Access & Settings
- Assigning privileges to a user group requires the manage privileges privilege.
- The Configuration menu shows for any configuration read privilege. Its pages are renamed to match the menu ("General settings", "User groups", "External validation").
Sinks
- MQTT sinks have a Retain setting, off by default.
Diagnostics & Operations
- Bug reporting can be switched off locally on the settings page. It runs only when both this setting and the Factry portal allow it.
- Session recording for bug reports has three modes: Never, On request (default) and Always. On request keeps the last minute in the browser and sends it only when the user attaches it to a bug report. The portal can lower this setting but never raise it.
- A bug report sends what the reporter enters (a message, their name, prefilled with their username, and an optional email address) and the recording if they attach one. Recordings mask all text and block media. Browser cookies are never sent, and a server that cannot reach the portal sends no error reports.
- New ARROW_COMPRESSION_CODEC (none, lz4, zstd, default zstd) sets the compression of Arrow and Grafana query responses. lz4 uses much less CPU and memory on busy servers, but compresses less, so responses are larger and use more network bandwidth. A client can also pick the codec per request.
- New EXPORT_PATH restricts Parquet exports to one folder. The server warns at startup while it is unset.
Opt-in Features
These features are off by default. Turn them on by adding their name to FEATURE_TOGGLES_ENABLE (comma-separated) and restarting Historian.
- named-events: events take a name from their trigger value, and events with different names can overlap on one asset. Child events can link to their parent by name, and manual events can be named and renamed. The events overview shows the name and can filter on it. A new From child events detection mode builds one parent event per name from the events below it (for example a batch from its unit batches), with a gap tolerance.
- lookuptablesync: a lookup table can be synced from an external database query, on an interval or with Sync now, with a preview of the query. Previewing and syncing also need READ_EXTERNAL_DATABASES. An optional watermark query skips a run when the source has not changed. A failed sync keeps the existing rows and shows on the homepage issues panel.
Bug fixes
- Fixed several places where a request could read or change data of another organization, including collector point writes, measurement reads by a failover collector, forwarders, event property values and log filters.
- The API enforces the Editable setting of manual entry form properties, not only the interface.
- Manual entry forms no longer freeze when the timestamp of a new periodic value is cleared, and no longer preselect 0 in number dropdowns.
- Restarting an event configuration no longer leaves its old scripts running.
- Event configurations and property configurations can no longer be moved to another asset, event type or property, which left values on the wrong events.
- Recursively deleting an event configuration closes its open events.
- A recalculation retries when an input query fails, and is marked failed instead of done. A calculation whose input was deleted is held with a visible reason.
- Writes and deletes to InfluxDB 1.x time out after 20 seconds instead of blocking forever when the database stops answering.
- Arrow query results and Parquet exports include the measurement UUID and group-by tags, and have the same columns on InfluxDB 1.x and InfluxDB 2.x.
- Points with timestamps the time series database cannot store (before 1677 or after 2262, such as the OPC UA minimum date) are rejected instead of dropped or stored at a wrong time.
- Delete checks, "Referenced by" and prototype instances only match resources of the same organization and type. Deleting several measurements no longer leaves an asset property pointing at a deleted one.
- Updating a form, event type, collector, prototype or task scheduler that still references a deleted resource no longer fails with "not found in organization" (regression in v8.2.2).
- Asset prototypes no longer clear a parent asset that is not at the root of the tree. Measurement prototypes show their label names again and save label edits. Prototype lists honour IncludeInstances again.
- Asset Excel import errors are listed per row.
- External database queries read decimals correctly, including NULL, HANA and ASE decimals, and a query that fails partway is reported as failed.
- Saving a collector no longer cancels its pending update. A deleted collector returns not found and can no longer be reactivated.
- Sinks no longer lose or resend points when the destination, such as an MQTT broker, is slow.
- An invalid authentication provider configuration no longer logs out all of its users.
- Restoring a deleted organization no longer hangs.
- A scheduled export with an end date or run count and an end time offset no longer crashes the server on its last run.
- Clicking a calculation input, a prototype in the bulk update list or a worker pool task's delete button no longer throws an error.
- Testing time series databases continues after one fails or hangs.
- Passwords longer than 72 bytes, the context of a calculation whose script does not compile, an empty license token and similar invalid input return a clear 400 instead of a server error.
⚠️ Before You Upgrade
- The upgrade does not start while an event type has a property named event_name. Rename that property first. Deleted event types and properties do not count. New properties cannot use the reserved view column names (event_name, duration, event_uuid, start_time, status, source, ...).
- Event property values move to a new table. The per-event-type SQL views gain an event_name column, event_properties becomes a read-only view, and the old table is kept as event_properties_legacy. Values stored under a name that no longer matches a property (left by an earlier rename) are not moved: they stay only in event_properties_legacy, and the upgrade log warns how many there are. Values of a deleted property are moved, and stay hidden like the property. Direct SQL that writes to event_properties stops working. Views you built on event_properties keep reading event_properties_legacy, so they miss later changes: the upgrade log names them, create them again on event_properties. Views you built on v_events, v_assets or the per-event-type views are kept, and the upgrade log names any it has to drop. The backfill can take a while on large databases. The new table and its indexes take 5 to 6 times the disk space of the old table, which is kept, so check that there is enough free disk space before upgrading.
- Privileges:
- Saving or running a calculation script requires READ_LOOKUP_TABLES. The upgrade grants it to user groups (not collector groups) that hold manage measurements, manage event configurations or manage prototypes. Groups created later need it added by hand.
- Groups with manage user groups but not manage privileges can no longer assign or remove group privileges.
- REST API clients:
- Mis-cased JSON body keys are ignored: a required field gets a 400, an optional one is silently dropped. Opt out with HTTP_JSON_CASE_SENSITIVE=false.
- Bodies over 64 MB get a 413. Raise the limit with HTTP_MAX_BODY_SIZE. Product updates (256 MB) and Excel imports (101 MB) have their own limits.
- Validation errors are application/problem+json, with one entry per violation in errors. The error and info members are still present.
- Stricter validation returns 400 where v8.2.x accepted the request. OPENAPI_VALIDATION_ENABLED=false bypasses most of it, but not these:
- A time series query needs Start and a measurement selector (before, a query without Start read all history).
- POST /api/logs/query takes the optional Limit and Page as JSON numbers instead of strings.
- POST /api/timeseries/export needs CompressionCodec, CompressionLevel and Interval (at least 1s).
- Changing the asset or event type of an event configuration, or the event configuration or property of a property configuration, is rejected.
- GET /api/collectors/{collectorUUID}/measurements and GET /api/product-updates/{uuid} are removed. Use GET /api/measurements?CollectorUUID=... for a collector's measurements.
- Writes to a manual entry form property that is not editable get a 403.
- Bug reporting: the upgrade turns bug reporting on and sets session recording to On request. Recording only starts once Factry enables it in the portal. Set it to Never if recordings must never be made.
- We recommend setting EXPORT_PATH so Parquet exports cannot write outside one folder.
v8.2.4
Released on 28/10/2026
Bug fixes
- Calculations, forwarders and collector HA settings can no longer use measurements, assets or HA settings of another organization
- Testing a new external database connection requires the manage external databases privilege
- Force deleting measurements in bulk or through a collector checks the manage measurements and manage assets privileges
- The last user administrator can no longer be removed by deleting the user, changing group members or privileges, or an SSO login
- User groups can no longer be changed to the collector type through the API
- Secure collector settings changed for several collectors at once are stored encrypted
- An empty bulk event selection is rejected instead of deleting or updating every event
- Manual entries are no longer lost or reverted while the event is being processed, and an event no longer stays incomplete after all manual values were entered
- Events created through a manual entry form or the API must match the asset and event type of their event configuration
- A manual entry form that may not edit or view an event returns 403 or 400 instead of 500, and the form history shows the operator again
- Manual entry form cells keep their modal open during auto refresh, and the page stops refreshing after you leave it
- A rejected event or property configuration update, for example a script that does not compile, returns 400 and no longer stops event processing
- Moving an asset under itself or one of its descendants is rejected
- Query responses, calculation tests and the calculation context return NaN and Inf values as null instead of failing or returning an empty response
- Time weighted average queries with a period of 0, or that need too many windows, are rejected instead of running forever
- BUFFER_MIN_FREE_DISK_SPACE is applied (the misspelled BUFFER_MIN_FREE_DISK_PACE still works), and changing the HA settings of one collector also updates its paired collector
- Measurement, collector and audit trail details no longer reload every 10 seconds, the event configuration detail no longer repeats its error toast, and external database settings no longer crash on a new prototype without metadata
v8.2.3
Released on 29/09/2026
Bug fixes
- The API only reads a request body after the request is authenticated, and validation errors no longer echo submitted values
- Invalid input is rejected instead of stored for forwarder settings, manual entry form settings, metadata, task schedulers, collector registration and bulk status updates
- Recurrence rules that never produce an occurrence, or do so too rarely, are rejected instead of hanging requests, schedules or server start
- A forwarder with a reload interval below 1 second no longer stops the server from starting
- Updating the root organization no longer clears its root flag, and the root organization can no longer be deleted
- Number metadata values keep their full precision
- Timeseries value filters only accept the documented operators and conditions, and true or false on boolean measurements
- The list of values of an event property, used for event property variables in Grafana, applies the status, property and child asset filters again, and loads faster on large databases
- Deleting tasks accepts several statuses, types and worker pools again, and never deletes unfinished tasks
- Force deleting a measurement also deletes the indirect dependents listed in the confirmation, and deleted asset properties disappear right away
- Sampled calculations no longer report insufficient data after a timeseries write that partly failed
- Testing a calculation through the API with number or boolean test values no longer fails
- Parquet exports from InfluxDB 2.x no longer fail after 20 seconds, and a worker pool no longer stalls when a task finishes early
- A stopped MQTT forwarder is reported as stopped
- Users without the manage measurements privilege no longer get an error on the calculation detail page
- The bulk prototype update page no longer exhausts the database connection pool
⚠️ Before You Upgrade
- An event configuration of a manual entry form with an unknown EditableEvents value now denies event edits instead of allowing them.
- NewForwarder.Settings and MetadataProperty.DataType are now required in the API, and API clients that sent invalid values get a 400 or 422 where v8.2.2 accepted the request.
v8.2.2
Released on 16/09/2026
Bug fixes
- Requests can no longer read, change or delete data of another organization through references in the request body
- A manual entry submitted through a form can only write to the measurement of its own asset property
- Creating a setting now runs the same validation as updating one, and the license token can no longer be set that way
- A manual entry without an event now returns a validation error instead of failing the request
- Event queries with a duration filter given as a plain number of seconds work again, and a value that is not a duration is reported as invalid
- Timeseries queries from a Grafana datasource no longer fail when the query leaves its filters, tags or grouping empty
- Parquet exports of many measurements no longer run out of open files
- Parquet export files are smaller: file metadata is stored once, and the column encodings configured for the export are applied
- A parquet export that fails while writing its last file is now reported as failed instead of successful
v8.2.1
Released on 09/09/2026
Bug fixes
- Event queries from a Grafana datasource older than v4.0.0 no longer ignore the selected assets and event types
- Filtering events on a number property now compares at full precision, so close values are no longer treated as equal
- Prototype changes can no longer read, overwrite or delete rows of another organization
v8.2.0
Released on 01/09/2026
Overview
This release centers on a complete rebuild of the REST API onto an OpenAPI 3.0 contract, with broad work on stability and performance under load and refinements to event and calculation workflows. Users will mainly notice a faster asset tree, richer event detail, and a more informative homepage.
API
- Rebuilt the REST API on an OpenAPI 3.0 contract, with spec-based validation of requests and responses.
- Error responses now follow RFC 9457 Problem Details, with clearer messages and appropriate status codes (400 for invalid input, 404 for a missing measurement).
- Request validation can be disabled with OPENAPI_VALIDATION_ENABLED=false if a client sends payloads the stricter validation rejects.
- Added POST query endpoints for events, event deletion and logs, so filters too complex for a query string can be sent in a request body. The equivalent GET and DELETE endpoints are deprecated.
- The logs endpoints can now return each line's structured fields (tags), opt-in via a query parameter.
- Pagination query parameters were renamed to lowercase (page, limit, offset). Query parameter matching is case-insensitive, so existing clients keep working.
- The unused legacy GET /setting/ endpoint was removed.
Events & Calculations
- Event property fillers can now reference parent event property values.
- Calculations now automatically pause after repeated errors, or when their state cannot be encoded.
- Added first-datapoint handling for event triggers.
- Event type and property names are now length-capped to prevent database identifier overflow.
- Event detail now shows a clickable link to the parent event, showing the asset and event type instead of a UUID.
- A parent asset without its own event configurations now shows the re-evaluate button when its children have them.
- Event configurations can now be filtered by keyword, with a result limit.
- The global calculation context is now scoped per organization instead of being shared across the installation.
Assets, Prototypes & Navigation
- The asset tree now loads lazily, making navigation of large trees faster.
- Asset tree search was optimized on both the frontend and backend. Search now matches on a single character, includes children in its matches, loads lazily, and caps results at 250 to keep large trees responsive.
- Added a button to keep or update all conflicts at once in the prototype bulk-update modal.
- The asset path is now shown in the asset property reference view and on events.
- Expanding a large branch now shows a spinner instead of appearing to hang.
Performance & Stability
- Added admission control for user and system workloads, with configurable limits that cap query and request duration to keep the server responsive under heavy load.
- Improved performance of event querying and measurement processing.
- Large Excel asset imports are much faster: rows are streamed instead of loading the whole sheet, assets and properties are bulk-inserted in dependency order, and per-row refetches were dropped.
- Asset tree traversal and dependency resolution were sped up with dedicated indexes and batched queries.
- Time series concurrency settings now default from the available CPU count instead of fixed values.
- Cache statistics now report cold-cache misses and a per-query hit ratio.
Homepage & Monitoring
- Added an issues panel that surfaces current problems with your Historian installation.
- Added a clock and timezone indicator.
- The homepage now shows per-process CPU and memory usage.
- Log lines now carry their structured fields, shown in the log viewer next to the message. A field referencing a collector, measurement, event configuration, forwarder, time series database or organization is rendered as a clickable link showing the object's name instead of its UUID.
Collectors & Updates
- Collector updates now support changelogs.
- Product updates and collector versions can now be retracted: when Factry withdraws a release it is no longer offered as an upgrade and any pending collector update to that version is cancelled automatically, so a known-bad version can't be installed.
- Discovered measurements can now carry metadata.
- Added gRPC endpoints so a collector can read the assets it needs, and the collector name, owning collector UUID and measurement totals are now populated on existing gRPC responses.
- Stale collector statistics are no longer sent to OpenTelemetry and the Portal.
Interface
- Pressing ESC clears all on-screen notifications.
- Success notifications now disappear after 2 seconds instead of 7.
Bug fixes
- Fixed several places where a request could read or modify data belonging to another organization
- The collector measurement endpoints are now scoped to the caller: listing a collector's measurements is filtered by the request organization, and the collector sync endpoint only serves the collector the caller authenticated as.
- Excel asset import rejects duplicate rows and reports the correct created and updated counts.
- A user can no longer modify another user's API token.
- An event filter with an inverted time range (end before start) is now rejected instead of silently returning nothing.
- Measurement queries and cache refreshes use nanosecond precision, so points on the range boundary are no longer dropped or duplicated.
- Deleting time series with an open-ended range now evicts the cache correctly, and deletion progress is reported accurately.
- The collector write buffer recovers from corrupt or partially written queue files instead of failing to start.
- MQTT forwarding continues when a single point fails to encode, no longer crashes on a missing asset property or measurement, and testing an MQTT forwarder no longer fails with a type error.
- A forwarder no longer stops permanently after a database error.
- Shift-clicking now includes the full range in a bulk measurement selection
- A collector change that fails to save now returns an error instead of appearing to succeed.
- The audit log panel no longer hangs the homepage on load
- Excel export generates valid column names past column Z, and the measurement importer no longer fails on short rows.
- A new organization now inherits the root admin group mappings.
- Time-weighted average no longer fabricates windows before the first point, and returns an empty result for a series with no data.
- The bulk toolbar no longer disappears on the events page when the detail panel is opened.
- The license warning is displayed again, and the license error clears when switching organization.
⚠️ Before You Upgrade
- The REST API now validates requests against the OpenAPI 3.0 spec. Well-behaved clients are unaffected, but a client that sends malformed or unexpected fields may now get a 400 where it previously succeeded. If this blocks a client, set OPENAPI_VALIDATION_ENABLED=false to restore the previous behavior while you adjust it.
- The global calculation context is now per organization. An installation that relied on one shared context across organizations must define it in each organization that needs it.
- GET /api/collector/collectors/{collectorUUID}/measurements now returns 401 unless the caller authenticated as that collector. This route exists for the collector's own sync, so normal installations are unaffected. An integration that was calling it with a user or management token should use GET /api/collectors/{collectorUUID}/measurements instead, which returns the collector's full measurement list and has no incremental updated_after equivalent.
v8.1.9
Released on 26/08/2026
Bug fixes
- A manual entry form event property configured with 0 decimals now shows 0 when you reopen it instead of the default, so confirming the dialog no longer resets the setting
- The audit log list no longer fails with a 500 on large tables: a composite index on organization and timestamp replaces the sort that spilled gigabytes to disk
- Querying a string or boolean measurement in Arrow or Parquet format no longer returns empty values, on an InfluxDB 1 timeseries database; the same applies to tag values returned by a tag query
v8.1.8
Released on 24/08/2026
Bug fixes
- Re-applying an asset prototype no longer resumes a paused calculation: the status of an existing measurement is now kept, and only a genuinely new measurement defaults to Active
- Calculated event properties whose script fails now complete when the configured close-after has passed, instead of leaving the event open indefinitely
- Events whose simple manual properties already have a value no longer stay stuck in "incomplete" waiting for manual entry; a migration also releases events that were already stuck
- Manual periodic properties with a dimension are now taken into account when checking whether an event is waiting for manual entry, so those events can no longer close before the entry is provided
- Calculations no longer drift off their interval grid, both when a query is split by the maximum query period and when a fallback value is written for an all-bad input window
- Excel bulk import now rejects duplicate measurements and duplicate asset properties instead of importing them
Changes
- Upgraded to Go 1.27 and updated the dependencies that govulncheck flagged as reachable (gRPC, golang.org/x/text, excelize)
- Removed a request body dump middleware from the REST stack that had no effect
v8.1.7
Released on 15/07/2026
Bug fixes
- Empty bulk selections no longer match every measurement or collector in the organization: a selection with no UUIDs or an empty filter is now rejected; covers bulk delete, patch, timeseries delete and recalculate
- Bulk actions now run against a snapshot of the selection, so changing a filter mid-action can no longer widen or clear the target set; the bulk toolbar also no longer stays open with a stale count after a filter reset
- Deleting a forwarder no longer removes the other forwarders on the same sink and no longer panics on an unknown forwarder
- Fixed context leaks in several tasks (worker, calculation, event processor, deletion, export, forwarder) that could grow memory over time
- Reducing a sink's buffer size on a running buffer no longer deadlocks incoming writes
v8.1.6
Released on 02/07/2026
Bug fixes
- Made calculation scripts more robust against problematic use of the context:
- Scripts that build circular or very deeply nested values no longer crash the server; they now stop with an error instead
- Scripts that repeatedly grow values stored in the context no longer consume excessive CPU and memory
- Calculations sharing an array or map through the global context can no longer corrupt each other's values
- Fixed an issue with "historian advanced" event properties: Events could close too early with incomplete or empty values
- The timeseries cache could serve stale data as complete, making sampled calculations report gaps of insufficient data
- Nil or bad values on measurements with an array type now store (bad quality) instead of vanishing; parse-error count now shows in the response error
- Improved access control:
- Log queries without an explicit organization filter are now correctly restricted to your own organization
- Sentry could not retrieve its settings for signed-in users without the read-settings privilege
- Corrected permission checks for external databases, forwarders, sinks, timeseries databases, and users in the interface
- The installer now retrieves the service file path from systemd when upgrading
- Fixed the collectors bulk action toolbar not closing
- The measurement data panel no longer resets the selected tab after a periodic refresh
Changes
- A failed service file update during an upgrade is now logged as a warning instead of stopping the upgrade
v8.1.5
Released on 10/06/2026
Bug fixes
- Fixed log limiter creating a spurious suppression log entry when the limit is first reached
- Excel asset import now rejects UUIDs from other organizations or that don't exist, instead of silently mishandling them
- Connection error messages no longer include database credentials
- Fixed filesystem IO counters always reporting zero in statistics
- Fixed asset creation from prototypes failing on non-HTTPS deployments
Changes
- Event compilation errors and invalid Excel rows in measurement imports now return 400 Bad Request instead of 500 Internal Server Error
- Improved error message for event property calculation priority conflicts
- Updated authentication providers to use secure settings for storing their configuration, and added a migration to move existing configurations to secure settings
- CPU statistics now share a single sample window across all collection paths, preventing inconsistent readings between metrics
- Service file now correctly sets WorkingDirectory
v8.1.4
Released on 07/05/2026
Bug fixes
- Recursive deletion of measurements/calculations now works when using bulk selection
- Bulk instantiation and updates of prototypes now work when the prototype has a large number of instances
- Logging in with a user in the root organization now works without needing to be in a user group with the READ-SETTINGS privilege
- The first data reference in an event property calculation configuration can now be removed
- Event property configurations can no longer be created through the API using event type properties from event types other than the one configured on the event configuration
- Newly created calculations can now be recalculated without first refreshing the page
- MEF view now works when no assets are selected while using the tree view
Changes
- Improved handling of the selectedUUID URL parameter for measurements/calculations when the selected item is not present on the page
- Calculation and Excel parsing endpoints now return a bad request status instead of an internal server error when the error is caused by invalid user input
v8.1.3
Released on 20/04/2026
Bug fixes
- Fixed problem where recursively deleting a collector did not remove all of its measurements
- Resolved an issue where recursive collector deletion could accidentally remove more measurements than requested
v8.1.2
Released on 17/04/2026
Bug fixes
- Fixed a performance issue in sampled calculations doing more queries than necessary
- Added a migration to update any event properties in the database with a nil value causing them to the related event to not show up in the event views in the database
- Fixed issue with built-in bug reporting
v8.1.1
Released on 14/04/2026
Bug fixes
- Fixed a system task not starting, most noticeable when no more statistics were getting sent to the portal
- Fixed various race conditions in the event processor
- prevent premature property completion when stop time is still nil
- prevent different workers in the event processor overriding each other's changes
- Fixed close after not working for event property calculations, keeping them open indefinitely
- Ensure event properties column is always properly initialized when the event has no properties, preventing them from not showing up when using the event views in the database
- Resolve issue with frontend continuously calling endpoints when the user detail panel was opened
- Fixed an issue when adding a newly created user to multiple user groups
- Clear measurement table selection after performing a bulk update
- Fixed issues with asset import when re-creating deleted assets
Changes
- Improved performance of asset import by eliminating unnecessary database queries
- Improved performance of event re-evaluation
- Optimized processing of events with a lot a properties configured
v8.1.0
Released on 31/03/2026
Overview
This release focuses on streamlining event workflows, improving calculation robustness, and increasing overall platform consistency. It also introduces enhancements to resource management, data export, observability, and deployment flexibility.
Events & Manual Entry
- Added bulk deletion and bulk re-evaluation actions to the events overview.
- Manual entry forms now support:
- Sticky form groups that remain visible while scrolling.
- The ability to immediately enter event properties after creating an event.
- Editing event start and stop times via the bulk edit properties modal.
Calculations & Processing
- Changing a calculation type now automatically resets its calculation context.
- Added support for linear fill type in TWA aggregation.
- Enhanced sampled calculations so they continue advancing even when all input datapoints have bad quality.
- Improved handling of Tengo errors.
- Prototype instance overrides now default to resolution type “keep” when prototypes are updated.
Resource Management & Data Operations
- Added support for recursive deletion of resources with dependencies, including a confirmation screen listing impacted items.
- Introduced a configurable task cleaner to automatically remove old tasks from the database.
- Time series data can now be queried using asset property UUIDs via the REST API.
- Improved API handling of requests from deleted collectors, now returning appropriate responses.
Asset Selection & Export
- Introduced a new asset tree selection component, now consistently used in:
- Excel export
- Parquet export
- forwarder configuration
- Excel export now supports selecting specific assets before generating exports.
Platform, UI & Infrastructure
- Added an option to disable TLS encryption for gRPC communication between Historian and collectors, enabling deployments behind reverse proxies that handle TLS termination.
- Reduced startup log verbosity to improve log clarity.
- Replaced toggle switches with checkboxes across the UI for more consistent and explicit boolean settings.
- Made identifiers such as UUIDs easily copyable throughout the frontend.
- Added internal telemetry and error-monitoring integrations to improve observability and operational insight.
v8.0.9
Released on 30/03/2026
Bug fixes
- Fixed issue where open events without stop time were not being processed
- Fixed bug where sink buffer settings were not being saved
- Fixed log levels not refreshing the logs page and resetting filters back to presets
- Resolved issue where re-adding removed assets via Excel import failed
- Fixed datetime picker not always working the parquet export page
- Solved problem where measurements in the bulk update prototypes modal were wrongly being reported as non-existent
v8.0.8
Released on 12/03/2026
Bug fixes
- Fixed issue with file paths for collector updates
v8.0.7
Released on 11/03/2026
Changes
- Added template link to measurement bulk import modal
Bug fixes
- Fixed measurement bulk import failing on empty columns
- Fixed pagination in audit trails component
- Bulk selection now remembers selection across pages
- Fixed issue in bulk measurement excel import returning errors for valid data in some cases
- Resolved problem in collector settings migration where secure settings were not properly encrypted and saved
v8.0.6
Released on 05/03/2026
Bug fixes
- Fixed some tasks creating a lot of audit logs
- Resolved an issue in the event processor to avoid potentially slow restarts of workers
- Resolved a database error occurring when a collector migrates measurement settings in bulk
v8.0.5
Released on 12/02/2026
Bug fixes
- Fixed format issue in arrow result of influxdb 1 queries when returning multiple different datatypes
- Resolved additional issues with asset property template variables not being resolved correctly in prototypes
- Added missing input fields ('close after' and 'preview enabled') back for event property configuration of detection mode 'External database'
- Resolved database issue with large imports of measurements from Excel
- Fixed tags and timestamp not resetting in after being overridden with the helper functions in a previous iteration of the calculation
Changes
- Also allow to filter prototype based views by conflicts during bulk prototype updates
v8.0.4
Released on 23/01/2026
Bug fixes
- Prevent base64 encoding of string values for external database event properties query results
- Correctly determine event duration used in event property calculations
- Allow negative values in duration input fields for 'historian' and 'historian_advanced' event property configurations
- Fixed asset property template variables not being resolved correctly in prototypes
- Added missing event type property dependencies for event property calculations
- Fixed pagination not working in the log viewer
- Fixed audit trail tab for authentication providers and organizations
v8.0.3
Released on 06/01/2026
Bug fixes
- The postgres version check during startup now correctly uses the admin database connection
- deleted asset properties no longer linger in the memory cache
v8.0.2
Released on 19/12/2025
Bug fixes
- Fixed an issue where the login cookie grew beyond the size limit when a user belonged to many user groups via LDAP group mappings
- Resolved incorrect rapport of CPU cores in some environments
v8.0.1
Released on 16/12/2025
Bug fixes
- Fixed issue where dependencies for assets and asset properties were not updated after bulk import via Excel
- Resolved LDAP login issue for existing users
- Fixed a potential issue where multiple event property workers could be started for the same event configuration
- Improved handling of sampled calculations when evaluating periods with bad data
- Admin-only users no longer have the page crash when viewing logs
- Made the bulk edit button visible again in manual entry forms for events
- Resolved issue where entering a manual entry for a periodic values could cause a crash in some situations
- Fixed scrollbar in log viewer resetting on refresh
- Fixed issue in calculation test modal where fetching sample values could fail in some edge cases
v8.0.0
Released on 21/11/2025
Overview
This release introduces a modernized UI, streamlined workflows, and improved performance across all core features. It also includes important under-the-hood updates to ensure long-term stability and compatibility. Some of the new features include:
- 🏠 A home screen with statistics
- 🧮 Calculated event properties
- ⚙️ Bulk operations for collectors, measurements and calculations
- ↪️ Extended parquet export functionality
- 📊 Swinging door compression and noise filtering in the collectors
- 🔍 And much more.. see below for more details
UI/UX & Productivity
- New Home Screen A redesigned home screen now shows key system info, organization stats, recent changes, and news updates in one place. This makes it easier to judge system health and understand what changed without digging through menus.
- Improved Layout & Navigation Measurements, calculations, and collectors now use table layouts with detail panes instead of pop-ups, making it faster to browse and edit without losing context. Tabs inside each pane keep information organized, and help texts moved to tooltips reduce clutter.
- Bulk Editing & Data Operations Selected fields for measurements, calculations, and collectors can now be edited in bulk, reducing repetitive work during large configuration updates. Bulk actions also extend to measurement data deletion, with the deletion task now capable of handling multiple measurements at once.
- Better Discoverability & Filtering Events can be filtered by status, columns in the event overview are configurable, and asset properties include a data preview for quicker validation. Date pickers now support presets, speeding up common time-range selections.
Events & Processing
- Calculated Event Properties A new calculation detection mode lets you compute event values dynamically from other properties—ideal for generating richer, more contextual events without extra data sources. An OnlyChanges option helps reduce noise by ignoring consecutive identical values.
- Event Configuration status New status controls let you pause and resume processing, and the expanded state/health view makes it clearer when something needs attention.
Parquet Export Enhancements
- More granular Parquet export settings for finer control over file generation.
- Support for exporting via asset tree selection, making it easier to target specific parts of the process.
- Added Hive-compatible partitioning (by database/measurement) to improve downstream analytics workflows.
Collector Improvements: Compression & Noise Filtering
- Introduced swinging door compression in collectors to reduce data volume without losing meaningful trends, improving storage efficiency and export performance.
- Added noise filtering, helping collectors ignore insignificant fluctuations so only relevant, stable values are stored.
⚠️ Before You Upgrade
Please review these important steps to ensure a smooth upgrade:
- Check Postgres version The minimum supported Postgres version has been changed from 11 to 12. Please ensure that your Postgres database is version 12 or higher before upgrading.
- Readonly postgres user per organization is no longer managed by Factry Historian. Use the custom Factry Grafana Data Source to query events or manage Postgres users manually.
- Removed automatic InfluxDB datasource creation and related settings. Use the Factry Historian Datasource plugin for Grafana instead
- Check Username Case-Sensitivity Usernames are now treated as case-insensitive, this means you'll have to delete any users who have different capitalization in their name. For instance if johnDoe and Johndoe exist, delete one of them. Please ensure that no users exist with a name that only differs by capitalization before upgrading.
Breaking API Changes
- Measurements API
- PUT /measurements replaces PATCH /measurements.
- PATCH /measurements still exists but now expects a different payload structure.
- Time series API
- DELETE /timeseries/delete is replaced by POST /timeseries/delete to reflect that the endpoint creates a deletion task rather than deleting synchronously.
- Logs API
- The following endpoints have been removed:
- GET /collectors//logs
- GET /measurements//logs
- GET /logs/historian
- Use GET /logs instead, and apply tag-based filters (e.g., by collector, measurement, event configuration, forwarder) to scope results. Logs from before the upgrade will not be visible when filtering on tags added in this release (e.g. log_type, organization_uuid, event_configuration_uuid, ...).
Deprecations (Plan Ahead)
- “raw” measurement datatype is being phased out. Referencing raw measurements is now no longer possible, the raw datatype may be removed in a future release.