# Paused cluster timeouts

Symptom: everything was working, now all connections time out with `ServerSelectionTimeoutError`. No config changed. The cluster is paused.

## Confirm

Look at the Atlas UI: the cluster tile shows Paused. Free tier clusters auto-pause after inactivity; any tier can be paused manually to save cost. A paused cluster accepts no connections, and the driver reports it as a timeout, which sends you chasing the IP access list for no reason.

## Fix

Resume the cluster (UI or `atlas clusters start`). It takes a few minutes to become reachable. For always-on workloads on the free tier, note that auto-pause will keep biting; either accept the wake-up latency or move to a tier without auto-pause.

## Verify

The cluster shows Active/Running and the startup ping succeeds. Add a cluster-state check to your runbook so the next timeout starts with "is it paused" instead of ending there.