FAQ: Simplified agent management and Smart Agent

[Aaron.Schifman]

Common questions and answers about simplified Agent Management with Smart Agent 

This FAQ articulates some of the most common questions and answers about simplified agent management with the Smart Agent and the enhanced user interface, along with the new CLI.

In this article... 

• Simplified agent management basics 
• Licenses and packages 
• Supported Environments 
• Requirements 
• Strategy
• Tooling Pipeline Guidelines 
• UI Questions 
• Installation/Getting Started 
• Custom configuration files, monitors, and extensions 
• Additional resources Updated 1/24/24

Simplified agent management basics 

FORUM LINK | Join the discussion on this topic

What is simplified agent management? 

Simplified agent management enables you to automate agent management operations such as installations, upgrades, and rollbacks at scale. This is achieved using the Smart Agent that facilitates and orchestrates agent operations by using the existing and enhanced Controller UI and/or command line interface (CLI). This applies to both SaaS and OnPrem customers alike.  

The Smart Agent can manage the supported agents when the Smart Agent is installed on the same machine as a supported agent. You can use the enhanced Controller UI to view all the installed agents with the inventory details. The Smart Agent inventory is also displayed along with other agent inventories. (See below for more details.) You also have access to a CLI for advanced management controls. 

Is the Smart Agent UI part of the current SaaS and On-prem UI, or will a separate installation be needed?  

As of Controller v23.11, the existing Agent Management screen has been modified to include new information regarding Smart Agent, whether agents are being managed, their status, and more. To access this new Smart Agent UI, you will need to be on v23.11 at minimum. 

What is the real value of this UI release? Will there be more than what we already have today? 

This release offers significant standalone value. However, using Smart Agent enhances this value by providing the following features: 

Simplified Upgrade Management	Enables you to choose the version to upgrade, download it from our repository, and make the necessary configuration changes all at the push of a button.
Upgrade Awareness	Effortlessly keep track of available upgrades.   These upgrades are visible on the right side of the screenshot, with distinct icons indicating their status—whether they're out of date, have updates ready for upgrade, are running the latest version, or have an unknown status.
Upgrade Rollback	Offers the ability to revert an upgrade if needed.
Progress Tracking	Provides a Tasks in Progress feature to keep you updated with ongoing processes.
Historical Logs	Grants access to a comprehensive history of logs for reference and troubleshooting.

Does the Management UI require the Smart Agent to be installed? 

No, Smart Agent does not have to be installed. However, the advantage of having the Smart Agent installed on the machine alongside the language agents is to have deeper insights and management capabilities.  

When Smart Agent is installed, you will be able to: 

• Inventory functionality, which in the initial release includes agent status (Out of Date, Update Available, Latest, Unknown), the agent’s version, running tasks, and historical events 
• Push-button management of upgrades and rollbacks on the machine 

Does deploying the Smart Agent require uninstalling existing agents, or is this an incremental process that builds from the existing installed base? 

This is an incremental process; you don't need to uninstall anything.  

Step 1: Install Smart Agent on the host.  
Step 2: Manage your existing and new language agents through the UI. 

Will the Smart Agent for agent management have an impact on host compute consumption? If so, what is the expected additional compute load? 

While there will be some processing that requires computing with regards to the Smart Agent being installed, it is thought to have no more impact than what an agent does today. 

Can users turn off agent management if they choose to do so? 

No. Because the Smart Agent is essential for maintaining agent lifecycle workflows, we haven't seen a significant need to develop this particular capability at this time.

Can Agent Management also be used to migrate from AppDynamics On-prem to SaaS

What if you need to upgrade an agent while also repointing it to communicate with a new SaaS Controller? 

Yes, you can use Agent Management to migrate from On-prem to SaaS with Ansible scripts. However, supporting this with a UI is under discussion. 

Does agent installation equate with app instrumentation? 

No, you need to restart the app for it to be instrumented.  

• The agent is automatically attached to the application in the cases of Tomcat, WebLogic, Glassfish, and Sprint 
• In other cases, you need to attach the agent for new installations as they did before Smart Agent by modifying the startup parameters. 

What will happen when an existing agent is running as a service?  

Existing agents running as a service will be backed up during the new agent version deployment and the same configuration will be used upon update. Nothing changes as to how the service is installed or is run. 

Back to TOC  

Licenses and Packages 

FORUM LINK | Join the discussion on this topic

Do I need to buy new licenses for this solution in addition to my existing licenses?

There is no additional cost to use Smart Agent. 

Is the solution license-aware in that it includes logic to prevent me from over-deploying agents compared to my entitlements?   

The current release of Smart Agent does not prevent the over-deployment of agents in relation to your entitlements. However, this functionality is under serious consideration for future updates. 

Is Smart Agent officially supported by Cisco AppDynamics if I have problems?  

Yes, Smart Agent is officially supported. 

How can you manage multiple license keys and utilize the charge-back license model? 

To manage multiple license keys for the charge-back license model, you will be able to specify a custom license key in the Additional Configuration section during deployment. 

Back to TOC 

Supported Environments 

FORUM LINK | Join the discussion on this topic

What agent flavors are supported? 

As of November 2023, the following six agents are supported: 

• Java 
• Node.js 
• Python 
• PHP 
• Apache Webserver 
• Machine Agent  

In upcoming releases, look for support for .NET Agent, DBMon and NetViz, among others. Check the documentation for the most current list of supported agents.

What features are supported in simplified agent management? 

Supported simplified agent management	Agent installs  Upgrades and rollback via the agent management UI & CLI
Management options	Bulk management options  Tasks in Progress and History tabs  Logs for troubleshooting
Agent Inventory	Running inventory of running agents with respective versions  Grid export – with end-to-end accounting of agents running and communicating with the Controller
Filtering	New filters include:  Out-dated agents  Available updates by application
Auto-attach for Java frameworks *	Auto-attach is supported for Java frameworks (Tomcat, Sprint, WebLogic, Glassfish) and Node.js. With it, admins only need to restart the application because the Smart Agent will know what startup parameters are needed. Otherwise, the startup script must be modified manually.

Are all the Java APM documented deployment scenarios (for different frameworks) supported? 

Yes 

Is auto-attach supported for Java deployments? 

Tomcat, WebLogic, Glassfish, and Spring are supported with this release, and many others are soon to come. Node.js is also supported. 

Does Smart Agent support multiple instances of language agents on a single node? 

Yes, each language will leverage an instance of a language agent, which will also be managed by the Smart Agent. 

Will the Agent Management UI support the ABAP Agent? 

As of the v23.11 release, the ABAP Agent is not included as part of the initial UI rollout. It is under consideration for the longer term.  

What about the DB Agent? 

Discussions to include the DB Agent are underway. We will apprise the Community of its support in a release. 

Are PSA and Machine Agent extensions also in the scope of agent management? 

PSA and Machine Agent extensions are part of the investigation phase, so there are no timelines to include them yet. 

Does Smart Agent support Docker applications (Java, .NET, Node.js, PHP) that are not running on Kubernetes?  

The Smart Agent CLI can be used in building Docker containers. It provides a uniform, automated way of installing App agents into containers during build times. 

Will I be able to use Smart Agent to manage agents that run on ECS Fargate or Docker containers?  

Yes, by integrating with Smart Agent CLI you can deploy agents during build time. 

Can we deploy custom extensions to all existing Machine Agents as mass deployment? 

Extensions are not supported as of this first release, but are part of future planning. 

How will Smart Agent be supported?

Regarding Support, we will follow the standard support process for agents, supporting agent releases one year after their release (n+1). 

Back to TOC

Requirements 

FORUM LINK | Join the discussion on this topic

Is a Smart Agent needed on every host? What are the various ways to deploy the Smart Agent? 

Yes, one Smart Agent is needed on every host that has applications participating in APM, as it will manage the application agents.

Installation methods include:

• manual installation
• use the AppDynamics-provided Ansible playbook meant for Smart Agent
• use existing agent playbooks
• use Smart Agent CLI for bulk rollouts of the Smart Agent

Are older agents removed when agents are upgraded? 

When agents are upgraded, one prior version of the agent is backed up for use in the event of a rollback.   

What network configurations and host permissions are required between host and Controller, as well as access to the internet? 

Network configurations and host permissions
Install time	Root permission: Most package managers have an implicit root permission requirement due to needing to write to /etc/environment. When a user has a package manager they use to install without root/sudo, then that will be supported.
Runtime	Depending on which agents are being managed, either root or an account with necessary permissions is required for runtime.  Neither Java nor Node.js require root during runtime.   Machine Agent, Python, PHP, and Webserver all require root, with or without the Smart Agent.
Network point of view	CSaaS: The port to connect the Smart Agent to is the same as the port other agents connect on. No new port is exposed. 443/80 are the ports for agents as well as Smart Agent now. On-Prem: A different port, -8030, is used for On-prem

What is the minimum language agent version required for Smart Agent to work?  

If you are running agents on a supported version of the technology, agent management through Smart Agent will just work.  

In the case of technologies that are End of Life (such as Java 6), Agent Management is not supported. Agent compatibility is the same regardless of how they are installed or managed.  

This is not a Controller release requirement because we are just managing agents.  

Is unidirectional or bi-directional network communication used with Smart Agent? 

Smart Agent uses unidirectional Agent-Controller network communications. 

Back to TOC

Strategy... 

FORUM LINK | Join the discussion on this topic

How does Smart Agent address managing hundreds of new and/or existing agents? 

Smart Agent is aimed at supporting existing customers with thousands of agents, without their having to uninstall or replace them. 
 

How does Smart Agent work?  

With the Smart Agent, you will automatically be shown all existing agents communicating with the existing Controller. So, there is nothing you need to do before gaining immediate value. Agents will appear on the UI and show whether or not the Smart Agent is also a part of that system.

If you add a Smart Agent to these machines, this enhancement gives you the additional options to install, upgrade, and rollback. 

Can you install and upgrade Smart Agent via Controller UI? 

While initial Smart Agent installation is not available via UI—once Smart Agent is installed, you can upgrade it via the Controller UI. Additionally, you can install the Smart Agent through the CLI. 

Is there a way to install Smart Agent at the centralized location and manage the agents remotely?  

No; Smart Agent should reside on every host that needs to be managed. However, with the CLI, one can manage Smart Agent tasks remotely. 

Would Simplified agent management deployments using Smart Agent still require application restarts for the newly deployed agent? 

Yes, Simplified agent management simply orchestrates the agent deployments, updates, and rollback operations. Since agents themselves require a restart, the intent for Agent management is to also manage the automation for that aspect but Smart Agent will not restart the applications themselves. 

Does Smart Agent have the option of deploying a custom customer agent build—for example on AIX-install that bundles customer JRE, etc.? 

Not in this first release. We are working towards testing this in the next release. 

What are the differences between the .NET Config Management tool, Smart Agent, and agent management? 

Agent management is a set of solutions we offer to customers to simplify agent lifecycle management. Smart Agent is a part of agent management that manages agents on the host.  

.NET Config management helps with config file management of .NET agents on machines. 

What will happen if the user has FSO Operator and Collector installed on a Cluster and the user is trying to install the Cluster Agent on the same cluster? 

The Cluster agent is outside the scope of this release but is being discussed for a later release. Should you need clustering support, the CLI can be used during build time. 

Back to TOC

Tooling Pipeline Guidelines 

FORUM LINK | Join the discussion on this topic

Does this support existing CI/CD pipelines? 

Yes, indeed...you can integrate into an existing CI/CD pipeline and are encouraged to do so.

Scripting is supported by this effort, and, with the Controller UI release, we are focusing more on supporting larger existing deployments with the capability of adding a Smart Agent to help manage upgrade, rollback, and deployment needs.

In a subsequent phase, users will be able to take full advantage of auto-discovery, where all they will do is install Smart Agent, and it will automatically detect which applications need instrumenting then apply the appropriate instrumentation based on those applications’ needs. 

How does the Agent Management's UI link to our requirement of using Ansible, Chef, Puppet, etc.? 

The scripts serve the CI/CD use case. Though the Agent Management console will use Ansible scripts under the hood, users will not need to engage with scripts. 

What prerequisites do users need to adapt?

For example, must they have Ansible Tower, a specific PoP, or have used Ansible scripts? 

To get started, install a single Smart Agent on each respective host as a one-time activity. After this manual or scripted installation, subsequent maintenance of the Smart Agent can be accomplished from the UI through Upgrades. 

There is no Ansible requirement. 

Back to TOC

Agent Management User Interface 

FORUM LINK | Join the discussion on this topic

How do I know whether the Smart Agent is installed? 

If the Smart Agent is installed, you can view it on the Controller Agent Management console/UI. 

Can we to schedule upgrades? 

Though upgrade scheduling is not part of this release, there are plans to provide this functionality in a later release. 

What about RBAC?  

Role-based access permissions that can be assigned to users will be available in the next release: 

• ITOPS Admin – can perform all the operations for all the applications. 
• AppOps Owner – has permission to perform actions to only their application. 
• DevOps – can view inventory for their applications but can’t perform any operations. 

Can data in the Agent Management Controller page be exported? 

Yes. You can export the data in the Agent Management Controller with the export grid function. 

Will we be able to modify app, tier, and node names for the agents in the Controller? 

Yes, you will be able to modify app, tier, and node names in the Controller. 

Does the User Interface include information that will help me understand when agents were installed or updated—for instance, a date or timestamp? 

Yes. 

Is there an option for bulk installing and upgrading Java Agent, Machine Agent, EUM, and others together? 

You cannot install different agents at the same time, nor can you perform bulk upgrades of different types at the same time. You can only install or upgrade one type of agent in bulk, like all Java agents. 

Can we monitor the uptime of Smart Agent as we do for the app agents? If so, can we create health rules on Smart Agent uptime? 

This release doesn’t have Smart Agent self-monitoring. However, we are planning to add this feature in the subsequent phases. 

Can we group agents by policy, in such a way as to enable one config change for multiple agents at once?  

For example, "AWS Linux 2.0" or "Application Servers"?

We are considering whether to tag the hosts in a later release.

In cases where app owners also own the servers, how can clashes between Controller and server admins be avoided? 

If the agent is currently being updated and someone tries to update the same agent while the upgrade is in progress, they will receive a warning notification. HTTP_PROXY can be set as an env variable in the Smart Agent, and it would be honored. 

How will Smart Agent work for applications like WebSphere, where the JA is passed via User Interface only, or via wrapper only?

You can provide values in the Additional Config section. 

Do we have an Auditing Capability? 

Not at this time. We intend to include auditing capability in an upcoming release. 

Is there a rollback-to-the-last-version capability in case a mass rollout causes unintended monitoring issues?

Yes, this is a part of the v23.11.0 release. 

What tools are available to support troubleshooting failed agent installations and upgrades?

Yes, readable deployment and upgrade logs are available in the Ansible if the Controller is using Ansible, or through the UI when using the UI with the Smart Agent. 

For SaaS or On-prem cases, can I customize the location from which deployment images are sourced? What security will be placed around selected images and Image repositories?

An upcoming release will enable customers to stage deployment from a customized location, which should also support security concerns. 

Back to TOC

Can we use the same Smart Agent for Kubernetes deployments? 

No, Smart Agent cannot yet be used in Kubernetes deployments.  

However, Smart Agent CLI can be used in container deployments with Docker. It provides a uniform automated way of installing App agents into containers during build times. 

Can we upgrade two PHP agents running on the same server for different PHP runtime versions with Smart Agent? 

Yes. 

Back to TOC

Getting Started: Installation

FORUM LINK | Join the discussion on this topic

Would the customer be able to install the agents manually even if the Smart Agent is installed on those hosts? 

Yes. Since they are registered to the Controller upon installing them manually, the agents will then be able to be managed by the Smart Agent as well.  

Does the CLI tool use SSH? 

Yes, CLI uses SSH or WinRM in the case of remote host Smart Agent installations. 

Do you need to deploy an SSH certificate to use Smart Agent? 

No certificate exchange is involved with regard to Smart Agent. 

Does the Smart Agent require root access or the creation of a user or group? If so, which agents are affected? Does this user have sudo privileges?

Yes, it uses root. This is because some of the agents - Python, PHP, Apache, and Machine Agent are currently installed as root. So, for Smart Agent to manage these agents, we need root permissions.

If the host only has Java or Node.js Agents, then we need not have root access to run them.

During agent installation or upgrade using Smart Agent, does it also restart the applications? Or will there be an option to restart applications? 

No, application restarts are not included. 

Back to TOC

Custom configuration files, monitors, and extensions 

FORUM LINK | Join the discussion on this topic

During brownfield deployment, how are extensions or monitors going to be managed—for example, Analytics Agent as an extension to Machine Agent. 

Machine Agent extensions are carried forward to the next version during upgrades. This release does not offer management to extensions on their own.

The Analytics Agent is not supported in the v23.11.0 release. For the latest information, see the Supported Agents section in the documentation.

During a brownfield deployment, does Smart Agent retain existing custom config changes on the agent—correlation XML, ISDK deployed? 

Yes. 

Back to TOC

Additional resources 

In the Community

• Smart Agent lifecycle management reimagined
• Exploring an APM agent upgrade scenario with Smart Agent 
• Exploring an APM agent installation scenario with Smart Agent 
• Smart Agent: How easy is it? — a short, clickable demo
• ^NEW!Just one Smart Agent: unlimited agent lifecycle control — overview video (9 minutes)
• Smart Agent Forum — add your own comments and insights here

In the Documentation:

• Agent Management documentation 
• Smart Agent documentation 
• Smart Agent Quick Start Guide 
• Smart Agent Command Line Utility (CLI) documentation