Agent Client Collector upgrade overview

  • Release version: Zurich
  • Updated May 28, 2026
  • 3 minutes to read
    • Summarize
    Summarized using AI
    This content was generated using new OpenAI-powered functionality. Results are provided on an as is basis and are not guaranteed to be accurate or complete.

    Summary of Agent Client Collector upgrade overview

    The Agent Client Collector (ACC) Framework enables automated upgrades of ACC agents directly from your ServiceNow instance, eliminating manual upgrades on individual hosts. You can upgrade agents individually via the UI or upgrade multiple agents simultaneously using an auto-upgrade feature with controlled rollout.

    Show full answer Show less

    Prerequisites and Preparation

    • Configure the Agent Client Collector websocket server on the MID Server.
    • Ensure MID Server, MID Web Server, and MID Server websocket server are running.
    • Assign the agentclientcollectoradmin role to users performing upgrades.
    • On Windows, upgrades run under the Local SYSTEM account; on Linux and macOS, sudo permissions with rpm/dpkg or pkg are required.

    Upgrade Methods

    • Single agent upgrade: Upgrade one agent at a time via the agent record UI or select up to 50 agents from the list view for batch upgrades. Recommended for validating the upgrade process.
    • Mass upgrade: Automatically upgrade all eligible agents using a scheduled job or a background script. Both methods apply rate limiting to control upgrade pace; the background script triggers immediate upgrades.

    Upgrade Process Stages

    • InstanceVerification: Validates agent eligibility based on OS support, version, status, and MID Server connectivity.
    • AgentVerification: Agent checks permissions, disk space, and dependencies to confirm upgrade feasibility.
    • Upgrade: Agent downloads, installs the new version, restarts, and reports status back to the instance.

    Each stage logs records in the Agent Upgrade Histories table for tracking.

    Agent Eligibility for Auto-Upgrade

    • Agent status must be Up.
    • Agent version between 2.7.0 and below the target upgrade version.
    • Agent must not be flagged as duplicate.
    • Valid MID Server or Pod reference required.
    • Failed upgrade attempts must be below the retry limit (default 3).
    • Runs on a supported operating system.

    Handling Failed Upgrades

    An agent that exceeds the failed upgrade retry limit is excluded from mass upgrades but can still be selectively upgraded manually by an administrator. The retry count can be reset via the UI after successful upgrades.

    Package Download Sources

    • Agents connected via Cloud Services download upgrades from the ServiceNow CDN.
    • Agents connected via MID Server download directly from the MID Server; if unavailable, they fall back to the ServiceNow Install Server.
    • Custom download servers can be configured in the agent’s acc.yml file to override default sources.

    Rollout Strategy Recommendations

    • Start by upgrading a single agent via the UI to validate the upgrade process.
    • Upgrade 5–10 agents from different operating systems using the list view for broader testing.
    • Perform a full rollout using the scheduled job or background script for mass upgrade with rate limiting to minimize risk.

    The Agent Client Collector Framework manages agent upgrades directly from the instance, with no manual action required on individual agent hosts.

    You can upgrade Agent Client Collector agents one at a time through the UI, or upgrade all eligible agents at once using the auto-upgrade feature. After an upgrade starts, each agent downloads the new version, installs it, restarts, and reports the result back to the instance.

    Before you upgrade

    Configure the Agent Client Collector web server. For details, see Configure the websocket server on the MID Server.

    Ensure that the MID Server, MID Web Server, and the MID Server websocket server are running.

    Roles required: agent_client_collector_admin
    • In a Windows environment: Local SYSTEM account
    • In a Linux environment: sudo rpm/dpkg
    • In a macOS environment: sudo pkg

    Upgrade methods

    Two upgrade methods are available:

    Single agent upgrade
    Upgrade one agent at a time from the agent record in the UI. Use this method to validate the upgrade process before a broader rollout. You can also select multiple agents from the list view to upgrade up to 50 agents at a time.
    Mass upgrade
    Upgrade all eligible agents automatically using the built-in scheduled job or a background script. The scheduled job applies rate limiting to control the rollout pace. The background script also uses rate limiting, and triggers an immediate upgrade without waiting for the next scheduled run.

    Upgrade stages

    Both single and mass upgrade attempts, progress through the following stages. Each stage creates a record in the Agent Upgrade Histories table.

    InstanceVerification
    The instance confirms the agent is eligible for upgrade. Checks include operating system support, agent version, agent status, and MID Server connectivity.
    AgentVerification
    The agent confirms it can perform the upgrade. Checks include permissions, available disk space, and dependencies.
    Upgrade
    The agent downloads the package, installs it, and restarts with the new version.

    For details on the required properties for agent overview, see Agent Client Collector upgrade properties.

    High-volume upgrade does not support Agent Client Collector to MID Server communication via mTLS.

    When performing high-volume upgrade, all agents that aren't using the most up-to-date version are upgraded. No upgrade is performed on agents already using the upgraded version.

    No upgrade is performed on agents that are outside the application scope. For more information, see Application scope.

    An agent is excluded from high-volume upgrade when you reach the failed upgrade limit for an agent. The failed upgrade limit is specified in the sn_agent.auto_upgrade.retry_limit system property. The default value for this property is 3.
    • This property applies to both high-volume and selective upgrades that have failed.
    • When an agent reaches the failed upgrade limit, an Agent Client Collector administrator can still run selective upgrade on the agent.
    • You can manually reset the failed upgrade limit by selecting the Reset failed upgrade attempts option on the agent page (select All > Agent Client Collector > Agents and select an agent).
    • After a successful upgrade, the upgraded agent is considered to have zero failed upgrades (for future upgrades).

    Agent eligibility

    An agent is eligible for auto-upgrade when it meets all of the following conditions:

    • Status is Up.
    • Version is 2.7.0 or higher and lower than the target upgrade version.
    • Agent is not flagged as a duplicate.
    • Has a valid MID Server or Pod reference.
    • Has not exceeded the retry limit for failed upgrades.
    • Runs on a supported operating system.

    Package download sources

    During upgrade, each agent downloads the new package from one of the following sources, depending on its connection type:

    • Agents connected via Cloud Services (ICS) download from the ServiceNow CDN at https://cdn-install.sncapps.service-now.com.
    • Agents connected via MID Server download from the MID Server directly over the local network.
    • If the MID Server is unavailable, agents fall back to the Install Server at https://install.service-now.com. To enable agent fallback, ensure that the install server is specified on the ServiceNow instance.

    If your agents use a custom download server, configure the agent-upgrade-url-path key in the agent's acc.yml file. This setting takes priority over the default sources.

    Rollout strategy

    Use a phased approach to reduce risk when upgrading a large agent fleet:

    1. Upgrade one agent from the UI to confirm the process works end-to-end.
    2. Upgrade 5–10 agents from the list view, selecting agents across different operating systems.
    3. Perform a full rollout by enabling the scheduled job or running the background script.