New Features

New features in Stroom version 7.11.

Integrated Pipeline Construction and Translation Development (Stepping)

Improvements have been made in how to construct pipelines and step through data with them. A pipeline can either be opened directly from the explorer tree or via the large Enter Stepping Mode button from the Data Preview pane of the data browser.

Clicking the Enter Stepping Mode button also allows the user to create a new pipeline (by clicking New ) rather than opening an existing one if needed.

Stepping mode is used to develop/test a pipeline against some data and change the translation interactively. Once a pipeline is open, stepping mode can be toggled on and off by pressing the Enter Stepping Mode toggle button on the pipeline Structure tab. Opening a pipeline from the data browser button opens a pipeline and immediately enters Stepping Mode. In stepping mode the user will be able to select individual streams to step through from the meta list by clicking on the Source pipeline element.

When Stepping Mode is off the user can make any changes they wish to the pipeline structure or properties. Going back into stepping mode immediately applies any structure or property changes to the stepping process. The results of any changes made to pipeline structure and referenced docs can be viewed by pressing the Refresh Current Step button.

These improvements allow a user to change any aspect of a pipeline in one place and immediately see the impact applied to some test data without saving pipeline changes or navigating to another tab. The source data being tested in stepping mode can be changed at any time by selecting a different input stream on the Source tab. The user can change the filter on the Source pane to find different data to test.

When entering stepping mode from the Data Preview pane of the Data Browser Enter Stepping Mode button, only the data from that context is initially included in the filter. Users must change the filter to choose different data.

Ask Stroom AI

The new Ask Stroom AI feature provides information and AI insights into data contained in search result tables or any other tabular data in Stroom. Ask Stroom UI can be accessed from any table context menu or the toolbar wherever you see the Ask Stroom AI button.

One or more AI models must be configured to be able to use Ask Stroom AI. The models used must conform to the OpenAI API standard. Once configured users can ask questions such as “Summarise this table” and the Ask Stroom AI feature will provide AI insights for the provided data.

At present the Ask Stroom AI feature is a preview feature with many enhancements planned for subsequent versions of Stroom.

Dense Vector LLM (AI)

Dense vector index field support allows Stroom to perform approximation queries to find document containing similar words, similar meaning or sentiment, similar spelling plus errors etc.

Dense vector support leverages the AI model configuration provided for the Ask Stroom AI feature. Support is included for LLM embedded tokenisation and dense vector field storage in both Elastic and Lucene search indices. The Elastic and Lucene implementations are different and both are experimental in this release.

For the Lucene version you can now create a new field type in the UI and set it to be a dense vector field. The dense vector field has many configuration options to choose from such as the AI embedding model to use and limits on context size etc. Once an index is created with a dense vector field and the LLM has provided the tokens for the field data Stroom is able to search using the field.

When querying search terms are substituted for tokens using the LLM and then matched against the dense vector field. This mechanism allows for connections between words and tokens in both the data and the query to find documents approximately related to the search terms.

Plan B Histograms, Metrics, Sessions

Multiple improvements to Plan B to store:

  • Histogram data, e.g. counts of things within specific time periods.
  • Metric data, e.g. values at points in time, CPU Use %, Memory Use % etc.
  • Session data, recording start and end periods for certain events, e.g. user uses application X between time A and B.

Trace Log Support In Plan B

Plan B can now store trace logs supplied via pipelines. There are new additions to the Plan B schema to capture trace log data so it can be supplied to a Plan B store.

Improved Annotations

Annotations have been improved to add several new features and improvements:

  • Labels
  • Collections
  • Data Retention
  • Access permissions
  • Links between annotations
  • Recording table row details
  • Instant update of decorated tables after annotation edits

Pipeline Processing Dependencies and Delayed Execution

It is now possible to delay processing of event data until required reference data is present. For each dependency Stroom will delay processing until reference data arrives that is newer than the event data you want to process. Stroom needs to wait until reference data arrives that is newer than the event data it needs to process so that it knows it has definitely received effective reference data for the events.

Minimum Delay

Once reference data has been received and processed we can process the event data, however as reference data is cached we may want to add additional delay to the processing, this is achieved by setting a minimum delay. Even without any processing dependencies a minimum delay can be added so Stroom will wait the specified period of time before processing.

Maximum Delay

In some cases we reference data might be late, and we might not want to wait too long before processing. In these cases the user can set a maximum delay so that processing will occur after the specified time delay even without the dependant data being present.

Git Integration

You can now create Git controlled folders in Stroom that will synchronise all child content with a Git repository. Git content can be pulled and pushed if the correct credentials have been provided via the new Credentials feature.

See Git Repo for details.

Content Store

Stroom now has a mechanism for providing standard content from one or more user defined content stores. Content stores are lists of specific content items described in files with Git repositories. The content store shows a list of items in the user interface that the user can choose to add to their Stroom instance, e.g. standard pipelines, schemas and visualisations.

See Content Store for details.

Credentials

Many features in Stroom require credentials to be provided to authenticate connections to external services, e.g. Git repositories, OpenAI APIs, HTTPS connections. The credentials feature provides a location to store and manage the secrets required to authenticate these services.

Credentials can manage:

  • User name and passwords
  • API keys and tokens
  • SSH keys
  • Key stores and trust stores in JKS and PKCS12 format.

See Credentials for details.

Smaller Changes

  • Pipeline elements now have user friendly names and can be renamed without breaking pipelines.
  • Main Stroom tabs can be reordered by clicking and dragging.
  • Main Stroom tabs can be closed to the left or right of the selected tab via a context menu.
  • The underline selection colour of a Stroom tab can be customised by a configuration property.
  • Dashboards now allow individual panes to be maximised so that a table or visualisation can be viewed full screen.
  • Table cells, rows, columns, selected rows and entire tables, now have numerous copy and export options.

New XSLT functions

A number of new XSLT functions have been added:

  • add-meta(String key, String value) - Add meta to be written to output destination.
  • split-document(Sting doc, String segmentSize, String overlapSize) - Split a document for LLM tokenisation (experimental for Elastic dense vector indexing).

Dashboard & StroomQL Functions

A number of new dashboard and StroomQL functions have been added:

Create Annotation

Since annotation editing is now performed as a primary Stroom feature via a button and context menu on dashboard and query tables we no longer need to create or edit annotations via a hyperlink. There are some remaining use cases where users want to create and initialise some annotation values based on some table row content. The createAnnotation function can be used for this purpose and shows a hyperlink that will open the annotation edit screen pre-populated with the supplied values ready to create a new annotation.

The function takes the arguments:

createAnnotation(text, title, subject, status, assignedTo, comment, eventIdList)

Example:

createAnnotation('Create Annotation', 'My Annotation Title', ${SubjectField}, 'New', 'UserA', 'Look at this thing', '123:2,123444:3')

See createAnnotation for details.

HostAddress

Returns the host address (IP) for the given host string.

hostAddress(host)

Example

hostAddress('google.com')
> '142.251.29.102'

HostName

Returns the host name for the given host string.

hostName(host)

Example

hostName('142.251.29.102')
> 'google.com'

InRange

Returns true if the value is between lower and upper (inclusive). All parameters must be either numbers or ISO date strings.

  1. The input value to test
  2. The lower bound (inclusive)
  3. The upper bound (inclusive)
inRange(value, lower, upper)

Example

inRange(5, 2, 6)
> true
inRange(5, 5, 5)
> true
inRange(5, 6, 7)
> false
inRange(5, 3, 4)
> false

Stroom Proxy

API Key Verification

When you have more that one Stroom Proxy in a chain (e.g. Remote Stroom Proxy -> Local Stroom Proxy -> Stroom), the remote Stroom Proxy may receive requests from client using API Keys . On its own, Stroom Proxy is unable to authenticate an API Key as they were created by Stroom. Instead it needs to make a request to the downstream Stroom Proxy to authenticate the API Key. This Stroom Proxy can in turn make a similar request to Stroom.

Each Stroom Proxy is now able to cache and persist the verified keys so that it can perform the authentication without having to always make requests downstream. As it can now persist verified keys in a hashed form, it is able to perform authentication even when it is unable to contact its downstream host.

The following configuration controls this caching:

proxyConfig:
  downstreamHost:
    # How long to cache verified keys for in memory
    maxCachedKeyAge: "PT10M"
    # How long keys will be persisted for on disk in case the downstream
    # can't be connected to
    maxPersistedKeyAge: "P30D"
    # The delay to use after there is an error connecting to the downstream
    noFetchIntervalAfterFailure: "PT30S"
    # The hash algorithm used to hash persisted API keys.
    persistedKeysHashAlgorithm: "SHA2_512"
Last modified February 12, 2026: Update 7.11 change-log (e55baea)