MindAccess Developer Plan Subscribers and MindAccess Operator Plan Subscribers

What's new?

• Anomaly Detection Batch Mode: allows the user to train and detect anomalies with up to 1 Million datapoints for improved accuracy or covering a broader scope of data. This function is available in the same offering, identical API endpoint and it uses the identical algorithm (DBSCAN) in the background. The input data is pulled from IoT timeseries storage in the background. Processing is conducted asynchronous and can last up to 3 hours for complex calculations. The existing Analytics Service offering does not change, but compared to the interactive APIs the Batch Mode is charged as multiple API calls with one call equivalent to 20.000 datapoints. For example, training a model with 200.000 datapoints will be charged as 10 API calls. For the maximum amount of 1 Million datapoints, a maximum equivalent of 50 API calls will be charged per anomaly training or detection.

Fixed issues

• TrendPrediction: "propertyName" is set as mandatory parameter. There was a mismatch between the specification and the actual behavior for this parameter.

• EventAnalytics: corrected the yaml specification for /findTopEvents, as there was a mismatch between the specification and the actual behavior for a single output syntax.

• Spectrum Analysis: In some cases when the API has been inactive for a while, after requesting the API to process frequencies the customer gets an error code 500 "internal server error" which is caused by the inactivity due to internal server clean up processes. This fix changes the clean up internally, which means the API will still function even after periods of inactivity.

What's new?

• General stability and security enhancements.

• Simplified login behavior - when opening the passcode link from the CF CLI users will be now directly redirected to the Login form.

• General minor updates for Buildpacks and Backing Services.

Known issues

• Upgrading single Logme, MongoDB 3.2 and PostgreSQL 9.4 Backing Services instances to clustered instances is not possible.

Additional Information

• Cloud Foundry allows to create more Backing Services instances than included in the individual plans. A fix for this behavior is going to be implemented with the next release.

Known issues

• In the current version the logout with Internet Explorer 11 might not work. In this case please close the browser directly.

• After modifying the authorization of an user via Identity Management Service, the change is not immediately visible. The user must then close the browser and login again.

Known issues

• For security reasons a strict timeout is enforced. Currently this leads to the situation that the size of a file which can be uploaded/ downloaded depends on the network bandwidth at client side.

• When uploading files to the File Service, a file name must not contain blank spaces. Blank spaces in the file name will prevent the file form being stored.

• After deleting an asset, associated files might still be accessible for up to 30 minutes as this data is deleted asynchronously.

Known issues

• The maximum payload size of a message sent to Time Series Service as part of a PUT request cannot exceed 1 MB.

• The service will never return more than 2000 records in a response.

• After deleting an asset, respective time series data might still be accessible for up to 30 minutes as this data is deleted asynchronously.

Known issues

• All usages generated within a Cloud Foundry Environment cannot be collected by Usage Transparency Services. All usages from and to a Cloud Foundry Environment e.g. requests to APIs still will be collected.