Agent screen frozen or stuck loading: what to check first

An agent reports the screen is frozen, stuck on a spinner, or not advancing to the next call. It feels like a server problem, but it almost never is. A stuck agent screen is usually the agent's own network connection, their workstation, or — less often — the database server straining under load. Knowing that order saves you from rebooting a healthy dialer.

The agent screen is a web page that constantly polls the server. When a poll is slow or never returns, the screen appears to hang. So the question isn't "is the screen broken" but "where did the round trip stall."

Where a frozen screen comes from

Each click in the agent screen fires a back-end call. A freeze means one of those round trips didn't come back in time.

flowchart LR
  A[Agent clicks button] --> B[Browser sends AJAX]
  B --> C{Network path}
  C -->|Wifi drop or VPN| D[Request stalls]
  C -->|Reaches server| E{Web and DB respond}
  E -->|DB slow| F[Response delayed]
  E -->|Fast| G[Screen advances]
  D --> H[Screen looks frozen]
  F --> H

The freeze can happen at three points: the agent's link to the internet, the trip across to the server, or the server taking too long to answer because the database is overloaded. Two reports tell these apart.

What to check, in order

1. Is it one agent or all of them? One frozen agent points at that workstation or its connection. Everyone frozen at once points at the server or database.
2. Check the agent's network. Wifi drops, a flaky VPN, or a saturated home uplink all stall the polls. High Latency on the agent screen connection is the most common single cause.
3. Have them reload the page. If a fresh load is clean, it was a transient connection blip and the Agent session simply lost a poll or two.
4. Open the Agent Debug Log Report for that session. With agent debug logging enabled, it captures nearly every button click and back-end AJAX call. You can see the exact action that hung and whether the response ever returned.
5. Check for latency gaps. A missing stretch of latency entries while the agent should have been logged in means the screen lost contact with the server during that window.
6. If many agents stall together and the network is clean, look at database load. A slow DB delays every response and freezes the whole floor at once.

The Agent Debug Log Report is detailed — it can run to hundreds of entries a minute — so scope it tightly to the session and the minute the agent reported. Watch for a back-end call that fired with no response, or a long pause between an action and the AMI (Asterisk Manager Interface) event it triggered. That gap is your freeze, and where it sits tells you whether to blame the client or the server.

Start with the agent, not the box. Keep this next to the troubleshooting playbook, and when you've confirmed the screen is hanging on slow polls, dig into the numbers with the Agent Latency Report. VICIfast runs the box, watches DB load, and keeps the dialer healthy so a frozen screen is almost always the agent's end, not yours — see plans and pricing.