Why a Remote Agent takes up to a minute to deactivate

You open the Remote agent modify screen, change the Status from ACTIVE to INACTIVE, hit save, and then the phone rings again. And again. For almost a full minute. Nothing is broken — this delay is how VICIdial is designed to work, and understanding the reason makes it much less alarming the first time you see it.

The root cause: polling, not event-driven updates

The VICIdial dialer engine — the process responsible for deciding which agents get calls and when — does not watch the database for individual field changes in real time. Instead, it reads remote agent status on a periodic cycle. Between cycles, it works off its last known snapshot of which agents are ACTIVE.

When you set a remote agent to INACTIVE, you are writing a new value to the database. But the dialer does not see that write until it runs its next status check. That check can take up to one minute. During that window, the dialer still considers this agent available and may send it more calls.

This is the same pattern used across the VICIdial system for performance reasons — continuously querying the database for every eligibility decision on every agent would be expensive at scale. The Real-time report and the keepalive processes operate similarly, reading state on short intervals rather than reacting to every individual change.

The timeline after you save

sequenceDiagram
    participant A as Admin
    participant DB as Database
    participant D as Dialer engine
    participant P as Agent phone
    A->>DB: Set Status = INACTIVE (save)
    Note over D: Still on previous cycle snapshot
    D->>P: Sends call (still thinks agent is ACTIVE)
    P-->>D: Call connects
    Note over D: Next cycle — re-reads DB
    D->>DB: Query remote agent status
    DB-->>D: Status = INACTIVE
    Note over D: Agent removed from eligible pool
    D-->>P: No more calls routed

The worst-case scenario is that you deactivate the agent right after a cycle completes, and you have to wait nearly a full minute for the next one. In practice the delay is often shorter, but plan for up to 60 seconds.

What to do during the delay

Calls that are already in flight when you flip the status will complete normally — you cannot yank a connected call just by deactivating the remote agent record. That is correct behavior. The Campaign holds the call logic, and a call that has already been answered is independent of the agent's status field.

The Agent API also lets you perform hangup and call disposition actions on a Remote agent during a live session, which can be useful if you need to handle an in-progress call programmatically while the status change propagates.

The ra_call_control interface exposes current call state on remote agent sessions, which is useful for tooling that needs to know whether a call is genuinely active before taking an action.

For the full explanation of what the Status field does and how ACTIVE and INACTIVE affect call routing, see our complete guide to VICIdial remote agents. If you are just getting started with remote agents, what is a VICIdial remote agent is the right starting point.

Running VICIdial on a dedicated managed server keeps the dialer process stable and predictable. See VICIfast plans — provisioned in about 40 seconds, single-tenant, with no shared resources affecting your dialer's cycle times.