Keepalive

A keepalive is a heartbeat. It is a small, regular signal that says "I am still here" so the other side does not assume the connection died. You will hear the word in two different but related places: network connections that need to stay open, and background processes that need to stay running. Both rely on the same idea of repeating a check often enough to notice trouble quickly. Without a heartbeat, a system can look healthy right up until the moment you try to use it and find it has been dead for an hour.

On the network side, SIP (Session Initiation Protocol) phones send keepalive packets so that routers and firewalls keep the path open. This is closely tied to NAT traversal, because home and office routers tend to close idle connections after a short period of silence. Without a keepalive, your Phone registration can silently expire, and the first sign of trouble is a call that never reaches the phone even though the agent thinks they are logged in and ready.

Keepalive as a watchdog

In a VICIdial system, keepalive scripts run on a schedule to make sure the dialer's moving parts are alive. They check the queues, restart stalled processes, and rebuild files like the Conf file when Asterisk needs a fresh copy. If one of these checks fails, that is often the trigger for Failover to a backup. Think of the watchdog keepalive as the difference between finding a problem in seconds versus finding it from an angry customer call. These scripts are deliberately boring and repetitive, which is exactly what you want from something guarding your uptime.

The practical takeaway: keepalives should be frequent enough to catch failures fast, but not so chatty that they flood the network or the logs. When you tune a system, check that your keepalive interval matches your registration timers. A keepalive that fires after your registration already expired does no good at all. If agents complain that calls vanish for no reason, a misaligned keepalive is one of the first things worth checking.