Add RPA server to cluster
C:/RPA or INSTALL_DIR is the default RPA installation directory. Change it to your own if needed.
C:/RPA or INSTALL_DIR is the default RPA installation directory. Change it to your own if needed.
The instruction describes how to improve the RPA server performance by increasing the number of running bots.
To run RPA as a service without logging in Windows, you can use the AlwaysUp software. The software is not free and has a trial version for 30 days. With AlwaysUp, you can use RPA as a service in RDP session 0 in Windows. The approach was tested (RDP-in-RDP option, too), and no issues were found.
WorkFusion RPA has an RDP client that doesn't use Windows Credential Management. Earlier, the client featured a single fixed screen resolution that was impossible to change. However, in specific RPA scenarios, such as image-based RPA, Business Process (BP) implementation requires a resolution different from the provided one. Such inconsistency can lead to BP failures.
Some drivers and browser windows can remain open after the previous execution and affect the current processes.
The Master Agent, Unit Agent, and Workers connect via REST API during a reconnect procedure. The Worker application requests the Unit Agent in the RoboticsFlowPlugin plugin. The Unit Agent requests the Master Agent. To configure a reconnection timeout for a Worker and Unit Agent, follow the instructions below.
By default, ELK (Elasticsearch, Logstash, Kibana) has no checks for the RPA units. The statistics are available only in Bot Manager.
Sometimes, for an unknown reason, the Internet Explorer (IE) driver crashes while a session is created, which results in the RPA bot waiting for three hours. To avoid the long waiting, the rpa.timeouts.http.create-session timeout is introduced:
The article describes the root cause and the steps to resolve the Windows Server 2022 console crash after the special Spanish symbol ñ is copied to it.
In case you found no solution to your RPA issue, gather the following logs from the Master server and send to the WorkFusion Support Team for further analysis.
The how-to instruction is only valid for Work.AI versions up to 10.3.2. In 10.3.2, Python 3.7 was deprecated.
The article mainly refers to issues related to the crashes of such components as the manipulation framework, an Agent, or a Worker.
The article lists a number of the most common issues stemming from incorrect RPA configuration and tips to troubleshoot them.
The article relates only to Chrome, Edge, and Firefox drivers. Desktop and Internet Explorer (IE) drivers are delivered with fixed versions that you can't change. In case of problems with IE, try to change the browser version to the recommended one.
Custom code issues are too numerous and varied to cover in a single article. However, these are the basic steps to complete if you suspect your issue is due to the custom code implementation:
To investigate whether your issue is caused by an out-of-memory (OOM) error, act as follows:
To confirm and investigate an issue related to poor RPA server performance, act as follows:
The smart auto-reconnection feature allows you to log in to a specified Bot Unit via RDP using a Bot Unit account. Bot Master waits for each Bot Unit to disconnect a session and restarts the RDP process when a manual or tester session ends.
If a Worker process on Windows gets stuck, throwing the following exception489] - Error 5 reading HKEYPERFORMANCEDATA from the registry, the application stops working, and a manual restart is required.
RPA Unit disconnect
Bots are executed on a separate Java Virtual Machine (JVM). As a result, a stop or pause action for a Business Process (BP) cleans up all queues for this BP and does not allow the Worker to start processing a new task from the main Worker queue. However, it doesn't kill the Worker already processing the task from the BP, and execution of the Bot Task continues till the very end, even if you stop or pause the BP.
Cannot launch bots locally because of duplicated dependencies
This instruction describes how to troubleshoot the issue when the robot class code taken to a variable on top of the robotics-flow plugin cannot pick the dynamic variable value used in the plugin.