Cogynt v2.10 Release Notes
Cogility is delighted to announce that Cogynt version 2.10 is generally available for both AWS and GCP customers as of August 2023. Please see what's new below.
If you'd like more information, or want to have your Cogynt instance updated, please contact us.
Data Management Tool
- S3 connector region configurations are now pre-populated with all S3-supported regions.
Authoring
Importing and Exporting from Authoring
-
Projects are now exported with a .cef extension, while project components (partial exports) are exported as .pcef files.
-
Partially exported files can now only be of one type of component. For example, a selection of event types exported as a .pcef file cannot also be exported with an additional event pattern. Similarly, a group of event patterns cannot be exported with an additional event type that is not part of the exported patterns' dependencies.
-
When imported, .pcef files now present a preview of the contents of the partial file, allowing users the opportunity to view and confirm the contents in the file before proceeding with the import.
-
All imported files are imported as copies. In other words, if there is an ID or naming conflict in any of the components being imported, a copy is be created with the version number appended to the component name. (For example,
Event Pattern_1
.) -
A progress toast pops up during export or import to indicate that the action is in progress.
Windowing
-
Windowing is a new feature that enables computing aggregations of a given data stream in an event pattern over a moving window of time. Windowing serves to optimize state management in analysis by removing the state after an event has fallen outside of a defined window, allowing improved scalability.
-
Tumbling and sliding windows are supported.
-
Windows can be applied to both event time and real-time (Kafka Ingestion Time).
-
Event patterns with windowing can be partitioned over the window.
-
Lateness is implemented for event time windows to allow flexibility in out-of-order events.
-
Window Timeout is implemented to avoid stale, unresolved windows when data flow stagnates.
Refined Computations
-
Authoring now has a refined and improved list of computational functions that aims to removed redundancies while also providing additional useful functions.
-
Computations now support certain computational operations between sets/arrays and scalar values.
-
The input of a function can be toggled between set and scalar value to express the desired mode of operation of the function.
-
For the list of new and updated computations, refer to Built-In Computation Functions in the Authoring User Guide.
Time to Live (TTL)
- Time to live now only applies to events that have not triggered a solution in a pattern.
Output Windowing Renamed to Batch Output
- To avoid confusion with the newly introduced Windowing feature, the feature formerly known as "Output Windowing" is now called "Batch Output."
Provenance and Drilldown
-
Provenance data is now published to a unique topic per pattern with provenance enabled.
-
Previously, all provenance data was published to the
template_solution_events
topic. -
This change allows for more selective resetting of drilldown data.
Other Updates
-
Improved Schema Discovery loading times.
-
Checkpoint timeout and Savepoint timeout now share the same value in the Flink deployment configurations.
Workstation
Events
- The Event Details Viewer now also displays any collections that the event may have been linked to in a new Collections tab.
Collections
File Attachment to Overall Collections
-
Added the ability for users to attach files to an overall collection. Previously, this was only possible through attaching files to comments.
-
Users can preview attachments in-browser for applicable file types. The current list of preview-able file types are:
- application/json
- application/pdf
- audio/mpeg
- image/bmp
- image/gif
- image/jpeg
- image/png
- image/svg+xml
- image/webp
- text/plain
- video/webm
-
The file types that can be uploaded are:
- application/json
- application/pdf
- application/ppt
- application/pptx
- application/xls
- application/xlsx
- application/xml
- audio/mpeg
- audio/wav
- image/bmp
- image/gif
- image/jpeg
- image/png
- image/svg+xml
- image/webp
- text/csv
- text/doc
- text/docx
- text/plain
- text/rtf
- text/xml
- video/avi
- video/mov
- video/mp4
- video/webm
Comments and Replies
-
The UI has been redesigned to be consistent with the new attachments experience.
-
Added the ability for the user to sort comments by most recent or oldest first.
-
Deleting a comment that has replies now shows the parent comment as deleted, but still display any replies to it.
Collection Linking
- Added the ability for users to link collections as related with the option of providing a short description.
Added Locking to Collection Text and Rich Text Custom Fields
-
When a user is editing a custom field text or rich text field, the system locks the field from being edited by other users.
-
If a user is inactive for 30 minutes, their changes will be submitted and they will lose the lock so that other users can access the field.
Greater Control over Collection Sorting and Filtering
-
Added additional filtering to the Collections widget to allow users to specify which time field to utilize when filtering by time.
-
Added the ability for user to select "Updated At" as a time field to filter by. Before, only "Created At" was available.
-
Added an option to specify a sort order to the Collections widget to allow users more control over sorting by ascending or descending order.
-
Added tree view and search to Collection Custom Fields filter.
Collection Pagination
-
The collections widget now returns up to 100 collections per filter at a time and displays a count of how many collections are returned based on the filters/search performed.
-
A page selector allows users to view additional results if needed.
Collection Update Times
The update time of a collection is updated more frequently based on actions performed on a collection, including:
- Collection created or updated
- Comment created/updated/deleted
- Attachment created/updated/deleted
- Custom fields updated on the collection
- Items added/removed from the collection
- Collection linked or unlinked as related from the collection
Added Collection Update Logging to Kafka
-
Every update to a collection now creates a message in the Kakfa topic
_cogynt_collections
with a high-level schema of a collection and its details (excluding events and rich text custom field values). -
This data can be consumed for a variety of uses, including dashboarding or other applications.
-
It is highly recommended to set a retention policy on this topic, as it is expected to grow in volume quickly due to the number of messages processed in any given day.
UI/UX improvements
-
Checkbox Custom Fields on a collection now have a toggle for the "Show Checked" or "Show Unchecked" options.
-
Added the ability for users to search and view roles of users on the "Assigned To" user list.
-
Added the ability for users to search and view roles of users on the "Participants" user list.
-
Adding a URL on Rich Text Custom Fields no longer requires the user to enter
http://
orhttps://
.
Link Analysis
-
Added "Neighbors" feature. Left-click a node to highlight its immediate connections. Shift-click additional nodes to continuously add to the highlighted nodes.
-
Restored the "shortest path" feature: Left click one node, then shift-click a second node to show the shortest path between them. Only applies for when two nodes are clicked/shift-clicked. (Note: If there are multiple equivalent shortest paths, only one is shown with the dashed lines.)
-
Added a "Remove Selected" feature to the right-click context menu to allow removing nodes in bulk more quickly.
-
Added a "Group Selected" feature to combine selected nodes into a summary node.
-
Grouping a Link chart now groups nodes via a summary node that displays the count of items within the grouping. Double-clicking a grouping expands the grouping and double-clicking a second time collapses the group.
-
Removed the "Clear Manual Events" options from the context menu.
-
Changed the default layout to "Organic" and decreased the number of re-renders needed to improve the visual layout of the chart.
-
Changed query depth logic to be more consistent with the labels. Level 1 queries one level out, 2 does two, and so on.
-
Added a single halo to any nodes manually interacted with. A double halo appears for the last manually dropped node.
Notification Settings
-
Added a new option when creating a notification setting called "Process notifications for events already ingested."
- If checked, a notification is created for all events that have already been ingested as well as any new events or updates to events.
- If unchecked, notifications ignore any events already ingested and only process new notifications for new events or updates to events from the time the Notification Setting is created.
- The default behavior when a new notification setting is created is "unchecked."
-
Made a change so that Notifications that have been modified by users (Dismissed, Assigned To, Priority, Tag) do not get reset back to the "default" state when redeployments are made.
-
The Pause feature of notification settings has been removed due to complications with intended use.
Dev Delete
When deleting events in the Admin screen, the following items are no longer deleted:
- Collection Links: If an event was linked to a collection, then deleted and re-ingested, the event now still appears as linked to the collection if the ID of the event is unchanged.
- Event Detail Templates.
Other
-
Help text in widgets have been updated with links to documentation.
-
Confirmation modals (e.g., deleting an item) have been visually redesigned.
-
In the rare case that a Workstation user does not have a first name/last name, Workstation now attempts to show their email instead of a user ID in most places.
-
Controls on widgets are now hidden when a widget is not actively being used.
-
Custom field template names now have a 50 character limit to prevent UI unsightliness.
-
Added search capabilities to the filters that can become a long list:
- Notifications/Notifications Explorer widget: Tags and User filters
- Collections widget: Tags, Assigned To, and Participants filters
-
Report Builder PDF exports of collections now bold the names of custom fields to provide better visual separation.
Audit Viewer
-
Added logging for when files are uploaded and attached to a collection.
-
Added logging for when collections are linked as related to other collections.
-
Enhanced the user name search to also support free text search.
-
The collections audit data no longer includes the values of the comments, attachments, custom fields, and collection items due to performance impacts. However, these objects are still logged in Audit Viewer as their own audit objects as notes, attachments, custom field values, and collection items, respectively.
User Management
Changes to Super Admin Logic and Default Role
-
By default, the
super_admin
role is now the default role that users obtain when they log in. -
In rare situations where the
auth_provider
changes, the default role reverts tosuper_admin
to prevent a lockout scenario.
Other
-
Users can now be logged into multiple Cogynt environments in the same browser.
-
The "Notes Attachments" permission has been renamed to "Collection Attachments" due to the new capability allowing the attachment of files to a collection without needing to create a comment.
Known Issues
Deployment
- Resuming a paused deployment when a cluster has to spin up new nodes loses the save state.
Authoring/HCEP
Computations
-
The
toDatetime
function returns null when converting to epoch. -
Datetime millisecond resolution must be at most 3 digits long. More digits will cause deployed jobs to crash.
-
Computation migrations reverse any
list
function arguments to thejoin
function. To remedy this, remap the arguments to the correct order. -
The
product
function produces incorrect results when taking set of integers or floats as arguments. Avoid usingproduct
function with sets or arrays. -
A
product
between a set / array and a scalar value results in null outputs. -
The
getUrlPart
function causes deployments to crash. Avoid using this function. -
The
getHost
function returns an incorrect value whencidr
is provided as an optional argument. -
The
geoContains
function returns null. -
The
firstNonNull
function with a mixture of sets and scalar arguments crashes deployments. -
The
valueAt
returns incorrect results when using set of arrays as argument.
Data Management Tool
-
Data validation for file uploads is currently disabled.
-
Using a tab delimiter in S3 connectors results in null value outputs.
-
GeoJSON values published via the Manual Entry feature are incompatible with Authoring's GeoJSON format, and therefore cannot be computed.
Modeling
-
Connecting pattern nodes from one node to another fails to visibly create input out output elements in the designated event patterns, but the event type is secretly registered as valid elements in those patterns. The pattern is then effectively corrupted. If deleting the pattern is not an option, delete the connection between the two patterns, and select the correct option that will result in the deletion of the input or outcome element on the corrupted pattern.
-
Connecting patterns from the hierarchical view fails to make a connection from the group to the outcome. Users must select the outer group to make the connection icon appear in the top center of the group, then click and drag the icon to the outcome to form a connection.
-
Windowing settings are not copied when duplicating patterns.
-
Reference topics for windows that contain multiple partitions are unable to capture all events within the window. A workaround is to make sure the referenced topic has only a single partition.
-
The outcome selection option is disabled immediately after deleting an outcome from the pattern view. Refresh the browser page to re-enable the outcome selection option.
-
The
inside
andoutside
constraints between geo-coordinates and geo-polygons cause deployments to crash. Avoid where possible.
Deployments
-
Stopping a deployment sometimes causes the status of the deployment to remain
Stopped
without transitioning over toNot Deployed
. This inhibits redeploying the deployment, as the Deploy button remains disabled. If this happens, click the Stop button again to transition toNot Deployed
, which enables redeployment. -
Flink configurations using
ms
units fail to deploy. Useseconds
whenever possible. -
Imported projects do not preserve solution
COG_id
s from the original project, which causes existing results for any given partition in the project to fail to sync up with new results from the newly imported but identical project. -
Upon updating for the first time, any existing deployments that are
Running
appear as having aNot Deployed
status, even when they are actuallyRunning
.
Other
-
Deleting an event type immediately after it has been created, while in the event pattern canvas where the new event type can be viewed, can result in pattern data corruption and cause the pattern and event type to no longer be tractable.
-
Typing long names for deployment targets sometimes freezes the application.
-
Project "modified at" timestamps are updated upon load instead of updating only when modified.
-
The
contains in
andcontains
constraints between arrays and scalar values are reversed. -
Event type outcome names are not updated in the computations view or right-click menu in the pattern view when event type names are updated. To remedy this, delete the outcome and reconnect.
-
Pressing the ENTER key immediately after inputting a name in Schema Discovery causes the UI to crash. Refresh the browser to return to the previous view.
-
Connecting large amounts of fields in computations in quick succession may sometimes cause the UI to lag and connections to disappear. Connect at a moderate pace and pause in between connections to ensure each connection is saved.
-
Copies of a pattern may be exported alongside the original pattern.
Workstation
-
The Events Stream widget does not properly sort columns that are named with camel-case format or with spaced words.
-
When a Collections widget is first added to a view, the refresh and pagination buttons will not work until the user first clicks the 'Filters' button on the widget and clicks Apply Filters.
-
Occasionally, trying to clear the contents of a "text" Custom Field on a collection may not succeed when highlighting all content and using the Backspace key. A workaround is to clear the field by typing in a single space in the field.
-
In some circumstances, a retracted pattern solution on the Drilldown widget may change color between blue and gray. Retracted pattern solutions are intended to always be gray.
-
There are rare situations in which the Drilldown widget may show outdated information for an event. If this happens, a reset should be performed to force the reingestion of Drilldown data.
-
The + More button on the Event History widget may produce an error when clicked.
-
After performing Dev Delete in Workstation, the list of projects in the project dropdown in the Admin screen may display incorrectly until after refreshing Workstation. In some situations, a deleted project may still appear until another deployment in Authoring is performed.
-
The Export Builder (CSV mapper) currently does not respond to user click actions to expand elements.
-
After setting an event_types icon, it cannot be set back to "no icon".
-
The "Globe View" in the Map widget does not render properly.
User Management
-
The Add Groups button no longer appears.
-
At this time, using the
Default
role selector is not recommended. If the User Management application restarts, two default roles may be set, leading to user lock-out. This requires manual intervention from Cogility IT support.
Other
- Cogynt clusters with a license in the grace period incorrectly prevents users from accessing User Management and Audit Viewer.