10.0

Edit

Version 10.2.3

Release date

October 19, 2021

With the primary focus on testing and quality in Q3 2021, the release has resulted in nearly 200 bugs fixed. Major improvements to Intelligent Automation Cloud Enterprise include the following:

  • Improved approach to Data Purge with a possibility to clean up persistent entities for all services—Analytics, Workspace, Control Tower, and so on

  • Enhanced performance of the Optical Character Recognition component

  • Role-based access in WorkSpace to enhance information security

  • Improved WorkFusion Analytics to reduce its impact on the overall IA Cloud performance

  • New stay-on-single Data Store versioning strategy to accelerate Use Case delivery and cut data loss risks

  • Opportunity to choose from multiple IDPs to sign in to provide easy and seamless integration with customer’s user management systems

  • Support for MS SQL MultiSubnetFailover to enable faster connectivity recovery after failovers

Other critical improvements are aimed at enhancing infrastructure reliability and managing Secrets Vault Aliases via Control Tower.

WorkSpace

The release enables you to limit access to WorkSpace resources to users based on their roles. For more details, refer to the instruction.

WorkSpace performance is also enhanced due to the following:

  • Improved filter performance

  • Fixed Manual Task processing issues causing Business Processes (BP) to get stuck

Analytics

To enhance the WorkFusion Analytics performance, Tableau Refresh Extracts for all out-of-the-box dashboards are improved to be executed within 15 minutes. In addition, several ETL (extract, transform, load) improvements are implemented, including:

  • Eliminated CPU spikes at the beginning of the ETL process

  • Optimized scripts and indexes

  • Business Processes meeting 10k records per hour with enabled Analytics

The improvements reduce the impact of WorkFusion Analytics on the platform performance.

New stay-on-single Data Store versioning strategy for ODF 2

A new approach to version Data Stores is introduced to enable greater flexibility of ODF2-based Use Case development.

Before the modification, Data Store versions were identical to those of the associated Use Case. For every Use Case version, a separate Data Store set was created. The mechanism proved to be inefficient for specific Use Cases where the database schema wasn’t changed from version to version. The always-increment strategy forces engineers to migrate the Use Case operational data from one set of Data Stores to another. With frequent releases, the effort becomes time-consuming for developers. Besides, manual data migration is associated with high risks of data losses.

A new stay-on-single strategy is introduced to separate Data Store versioning from that of Use Cases. The separation allows reducing the total amount of data tables to handle while maintaining the possibility to version Data Stores. As a result of the improvement, customers can accelerate Use Case development and cut data loss risks and manual efforts within the process.

User management

The unified user management system was enhanced with the capability to choose between several Identity Providers (IDPs) when signing into the platform if multiple IDPs are integrated. Previously, users were logged in from the first of the numerous IDPs in the authentication flow.

The new feature facilitates the management of multiple IDPs, thus enabling customers to integrate their existing user management systems with the platform seamlessly.

Support for MS SQL MultiSubnetFailover

The support of MultiSubnetFailover is enabled for all database connections of the WorkFusion platform. The feature allows for faster reconnection after a failover, reducing the failover time for single and multi-subnet Always On topologies.

Improved Data Purge

The out-of-the-box Data Purge feature was improved to enable the cleansing of persistent entities across the WorkFusion services—Analytics, WorkSpace, Control Tower, and so on. Other enhancements included the following:

  • Data Purge was configured to run only within the preset purge duration.

  • Issues related to scheduling via Cron expressions were eliminated.

With obsolete data removed correctly and regularly, customers can benefit from the faster performance of the WorkFusion services and more efficient data management within the platform.

Updates

  • Reverted the update of Java to version 8u262 to eliminate the RPA component issue when the Worker failed to stop after stopping the Bot Unit in Bot Manager.

  • Updated Zulu builds of OpenJDK to version 8.56.0.21.

  • Upgraded Tomcat to version 8.5.

  • Updated Apache POI and Apache POI-OOXML to version 3.17 in BEP workers to unblock the development of RPA Use Cases with SAP integration.

  • Updated Kibana to version 7.11 to eliminate the Node.js 10.5.2 vulnerability to HTTP request smuggling.

  • Upgraded apr-utils to the latest available version:

    • For Centos 7—version apr-util-1.5.2-6

    • For Centos 8—version apr-util-1.6.1-6

  • Updated the following vulnerable libraries to the latest versions:

    • jQuery 1.8.3

    • Bootstrap 3.3.5

    • jQuery-ui-1.9.2

Improvements

Infrastructure

  • Updated the Mesos configuration to make its user interface disabled by default for security reasons.

  • Updated the RabbitMQ configuration to make its user interface disabled by default for security reasons.

  • Removed the user-creating feature from the installer to sustain consistent server configuration.

  • Enhanced the installer precheck operations verifying the correct location of mandatory Zookeeper’s datalog and optional Elasticsearch’s data-dir directories.

  • Improved installer’s port checking capabilities and reliability.

  • Stabilized the installer by adding prechecks to look for forbidden characters in passwords.

  • Updated the list of supported and forbidden characters for the Tableau passwords.

  • Modified the installer to eliminate the use of vault_keys.json. Now, the Vault root token is obtained from ZooKeeper.

  • Removed weak TLS 1.2 ciphers from the servers to improve the Product's security.

  • Enabled Metricbeat to deliver memory and CPU metrics from Mesos to ELK.

  • Enabled Metricbeat to deliver RabbitMQ metrics to ELK.

  • Enabled detection and notification about suboptimal or incorrect number of RabbitMQ nodes in clusters to improve reliability.

  • Added another validation step to reveal incorrect configuration on early installation stages, thus improving a troubleshooting process.

  • Stabilized the upgrade procedure by adding the validation of the latest components version to be installed.

  • Explicitly defined the list of Cross-Origin Resource Sharing (CORS) endpoints allowed to access Minio endpoints.

  • Removed useless flooding messages from Nginx logs.

  • Removed excessive jdk_win variable from sources.yml.

  • Removed excessive supervisord watcher from Elasticsearch, Logstash, Kibana (ELK).

  • Updated the BEP Cluster Utilization dashboard in Kibana with the two new charts—total CPU and RAM per host—to improve troubleshooting.

AutoML

  • Added support for ML-SDK archetypes with a new structure similar to quick-start archetypes.

  • Added support for checkbox detection in the model training flow.

Control Tower

  • Limited access to viewing the values for Secrets Vaults Aliases via the Control Tower user interface. Previously, users with the Manage Secrets Vault Entries permission could see the values during input or viewing. Now, the values are hidden behind asterisks. They become visible only after you click the eye icon.

  • Improved user data search by reducing the load time of Control Tower pages.

OCR

  • Improved the health check mechanism by adding the ocr.health.worker.version parameter that enables or disables a worker's health check per a worker type.

  • Updated OCR dashboards with the two new charts—document size and page count.

  • Optimized work with metrics in the High Availability (HA) mode. Previously, all nodes sent metrics to Logstash simultaneously. Now, the ZooKeeper lock is used to enable sending logs from only one node.

  • Reduced size of the OCR health check response.

User management

  • The loading speed of the Create filter page has been increased, thus enabling customers to create a filter faster after logging into Keycloak with LDAP credentials.

Bug fixes

ODF

  • For Bundle Maven plugin, connectionTimeout and readTimeout were made configurable to resolve issues that occurred when a project was deployed to UAT or test environments via the mvn bundle:import command.

RPA

  • Fixed a handle leak causing BEP Worker processes to get stuck. For that, a possibility was provided to configure the frequency of calls to MemoryUsageTracker. By default, MemoryUsageTracker gets system information every second. Two settings were introduced to adjust the intervals of getting memory usage statistics. In addition, users are now enabled to switch off memory usage tracking.

  • Enabled automatic restart of BEP Workers when they are forcefully closed during execution—upon the Ctrl+c signal, by pressing the [x] button, or due to a process killed from the console.

  • Fixed the EmailMessage class issue to enable the display of the received date for emails.

  • Eliminated a RabbitMQ startup failure due to an OpenSSL bug affecting users with processors supporting Intel SHA Extensions. As a workaround, the SHA extensions are disabled by setting the OPENSSL_ia32cap to ~0x20000000 environment variable.

  • Fixed the issue of execution failing with ClassCastException when RPAUtils.callMethodWithoutParams returns Object.

OCR

  • Fixed the memory consumption issue related to OCR JVM functioning as part of the OCR data purge.

  • Resolved the issue with the OCR ABBYY query blocking processes by fixing the Hibernate logic.

  • Eliminated the batch cleanup job issue.

  • Fixed the tag shifting issue.

  • Eliminated the orientation flag issue.

AutoML

  • Resolved the issue in connection with false Python representation of the LogisticRegressionCV parameters.

Control Tower

  • Fixed the Business Process issue associated with the "Failed to process event RunActionEvent" message dropped each second to Control Tower logs in Kibana.

  • Added the WARN logging mechanism for notifying about the duplication results received by Control Tower.

  • Resolved the user interface accessibility issues.

  • Eliminated the issue of records not getting deleted from a Data Store after it is renamed.

  • Fixed the issue of exceptions getting created in the log when the user tries to edit a Business Process while running.

  • Enabled switching to a different AutoML model when a user attempts to configure the AutoML tab after creating a Manual Task and deleting the hyper model from the S3 bucket.

  • Fixed the broken popup that appears when a user double-clicks a rule immediately after creating it.

  • Eliminated the issue of displaying an incorrect number of expired tasks on the View All Manual Tasks page.

Infrastructure

  • Fixed the RHEL 8/CentOS 8 precheck issues with incorrect verification of installed components.

  • Fixed the issue with the RabbitMQ Metricbeat module failing to start due to incorrect password characters.

  • Fixed intermittent Keycloak connectivity issues with the AD MS SQL login.

  • Fixed the issue with the secure.properties files deleted incorrectly after the installation.

  • Fixed the issue with the RabbitMQ HA nodes not recovering automatically after network problems.

  • Fixed the issue with the health check BP freezing on the Wait for Training Completion step during the upgrade procedure.

  • Fixed accidental issues with the installation failing because of the Nginx unsuccessful startup.

  • Fixed the issues with the Nginx error log due to enabled Online Certificate Status Protocol (OSCP) stapling.

Bot execution platform (BEP)

  • Fixed the issue with the /INSTALL_DIR/mesos/work/slaves/ folder on BEP Agents eating up space. Now, all temporal resources are cleaned up regularly.

  • Fixed the mismatch between the user and server time zones.

  • Enabled the Worker Management Service (WMS) to fix broken BEP Workers or activate the error mode when WMS finds such Workers.

  • Fixed the task duplication issue during task distribution in RabbitMQ.

  • Enabled stopping the BP after a predefined number of retries if there is an exception in the Rule step of the BP.

  • Fixed the 500 error that appears when a user tries to expand the list of BP steps on a BP list or opens the BP itself.

  • Fixed the issue with the freezing of the google-ads-java library.

  • Enabled the fair share rule for the distribution of stateless tasks.

  • Fixed the vulnerability of exposed actuator endpoints in BEP Agents.

  • Reconfigured the default base log directory. BEP Worker logs are no longer kept in a common Network File System (NFS) storage.

  • Fixed the issue of Bot Sources not getting released when an RPA bot is stopped.

← Version 10.2.4Version 10.2.2 →