If you run scheduled WebReports on Content Server, you have probably had this conversation: one WebReport needs to run at a precise time, another takes 20 minutes and holds up everything queued behind it, and a third appeared last quarter when a user gave their own WebReport a schedule, and nobody is entirely sure what it does. All of them share a single scheduler, and it treats them all the same way.
Enter the Ravenblack Multi-Agent Scheduler (RBMAS). As we are releasing a new major version of this product, we’re recapping what RBMAS does and how it augments WebReports Scheduling and previewing some of the features in this new release.
By default, Content Server runs all scheduled WebReports through a single, one-size-fits-all scheduling “agent” that manages one single queue of WebReports waiting to be run. Think of an agent as an assistant that has been delegated the task of running one or more WebReports at pre-designated times, for a specified number of repetitions. Every scheduled WebReport shares the same queue, runs on the same fixed timing, and waits its turn behind whatever else happens to be due—whether it’s a quick task or a long one, time-critical or not. Another aspect of this out-of-the-box functionality is that any WebReport developer can choose to schedule any WebReport to run at any time, further adding to the congestion on a single fixed queue of scheduled WebReports.
To address the limitations of the long-standing scheduling framework for WebReports, the Ravenblack Multi-Agent Scheduler provides several improvements, bringing WebReports scheduling technology up to modern day business requirements. In no particular order here are five key features of the Multi-Agent Scheduler:
Provides an approach to safely allow the separation of disparate tasks without conflict or confusion
This is the original marquee feature of RBMAS, allowing the easy creation of multiple agents and an easy mechanism to move WebReport tasks between agents.
Supplies uniquely configured (and named), purpose-built agents to suit different task types.
Because the current technology only allows one scheduling agent to be created, that agent must be engineered and configured to handle all possible use cases that a customer may require. The RBMAS provides a UI to create new agents with completely unique descriptive agent names alongside more tangible agent parameters such as the sleep timer, which affects how often the agent wakes up and runs.
Offers enhanced configuration options to allow more granular precision and event handling.
Given the new capability to define multiple agents with unique configurations, it made sense for us to start creating additional configuration options and additional unique behaviors and algorithms to match different customer use cases. Foremost among these enhancements is the ability to vastly improve precision for time sensitive business tasks, while also allowing some control over the scheduling algorithm to meet unique business requirements.
Performs detailed analysis and logging capability to allow full visibility and control of scheduling performance.
As a perfect companion to the uniquely configured, dedicated agents, RBMAS offers two powerful features for managing and monitoring the tasks running on each agent. The first feature enhances the ability to view scheduling task progress. Monitoring and debugging WebReports running autonomously can be a difficult process due to the volume of data and the number of WebReports generating logs to a single log file. With RBMAS you can move any WebReport onto a separate agent and then enable a separate “noise free” log file to capture all the task events for that agent. The second feature provides multiple unique measurement points to generate performance analysis information. This is particularly useful to compare and evaluate how different agent configurations have affected performance, and to identify potential issues before customers report them. Both features are complemented by our AI agent (RaiVE) that is explicitly trained to interpret these log files and provide recommendations and observations.
Presents improved UI features and enhanced admin functions.
The new RBMAS interface, and accompanying maintenance page deliver several important UI enhancements while largely replicating the original framework used by the existing scheduler page. Among the many improvements, RBMAS provides the ability to edit any schedule, even if it was created by a different user, in addition to: filtering, sorting, dynamic on/off toggling for features, a status indicator and a whole new page to manage system functions and to provide visibility of the registry that controls the multi-agent coordination.
It is fair to say that many customers have found alternative ways to address some of these issues, largely as workarounds using technology that is external to Content Server. These solutions are typically put in place to solve a single unique issue. While these solutions have enabled customers to solve their immediate scheduling issues, we believe that the RBMAS product offers a comprehensive, robust and secure solution to the standard WebReports scheduling technology. The RBMAS approach maintains all of the strengths and benefits of the core functionality by logically extending the original solution.
The five main features listed above have been core pillars of RBMAS since its initial design, but there is always room to continue expanding these features, and to adapt the product to meet changes in Content Server. The most recent version of RBMAS has introduced some great new features that we describe in more depth here. Hopefully with the detail provided here you can get a sense of the attention to detail we provide with our ongoing development program.
1. RaiVE - an AI extension that reads your scheduling data for you
This is the first time we’ve added our RaiVE assistant to this particular product. As with our other products it complements the existing functionality and takes advantage of the useful, curated data that RBMAS can generate, while helping administrators to summarize, process, and understand the information behind large volumes of data.
RBMAS's everyday scheduling agents already capture detailed, plain-language-friendly activity data for every WebReport they run. RaiVE can be pointed at one or more of the analysis logs, and it can interpret what happened—flagging things such as a WebReport that isn't keeping up with its own schedule, an agent that's carrying more load than it can handle, or a schedule that's been set up in a way that can never work—without anyone needing to read the raw logs themselves.
2. Use Precision - for reports that need to run right on time
Some reports need to start at a specific time, not "sometime in the next minute or two." The Use Precision feature lets an administrator enable much tighter timing for a specific agent, so it checks for “ready to run” WebReports at much shorter intervals, allowing them to start within a few seconds of their scheduled time, rather than continuously drifting. It's an option you can turn on for the reports that need it, without affecting anything else.
3. Overrun Handling - control over what happens when a report runs long
Every so often, a report takes so long to run that its next scheduled time arrives, or passes, before it's even finished. Left alone, that usually means the report picks right back up and runs again immediately, over and over, until its job is complete.
RBMAS now lets an administrator choose, per agent, how that situation is handled: either let a report pick right back up as soon as it can, or have RBMAS automatically skip ahead to the next time slot that hasn't already passed. This allows a slow-running report to settle back onto its intended rhythm rather than trying to catch up all at once. It's a small setting, but it hands administrators real control and choice over agent behaviour.
4. Distributed Agents - additional parallel processing options
For some years, Content Server has included a feature called Distributed Agents that allows many tasks to run in parallel across a pool of sub-agents. Starting with version 25.4, that capability became available for WebReports scheduling too—and RBMAS makes it easy to take advantage of, on one or more of the RBMAS agents. This new capability adds more choices for the administrator, and this is a useful option for low-priority use cases where precision and guaranteed single execution are not necessarily required. We also suggest this as a good option for the RBMAS default agent that runs the many ad-hoc WebReport schedules that can be created by any WebReports developer—effectively keeping them away from more formal requirements and processing them in an efficient way.
5. Dedicated Worker – further separation and priority for sensitive tasks
The Dedicated Worker feature lets an administrator assign any agent to a dedicated “worker thread”, rather than sharing one with every other RBMAS agent. This thread is responsible for starting up agents. Having a dedicated thread potentially protects an agent from being a victim of other tasks corrupting memory and crashing. A dedicated worker thread also ensures that an agent is not waiting on other agents to start up after a restart, which is important for time-sensitive agents. (Available from Content Server 24.4 onward.)
Whichever combination of these options a customer uses, RBMAS's everyday agents keep doing something Content Server's own scheduling tools don't: giving administrators clear, detailed insight into exactly what ran, when, and how it went. For more information on how RBMAS can help your Content Server team, reach out to us at info@ravenblackts.com.