When a node loses network connectivity — due to hardware link failures, carrier fluctuations, or configuration issues — it becomes unreachable for extended periods. A network interruption drill simulates this complete loss of connectivity on an Elastic Compute Service (ECS) instance to test whether your monitoring and recovery systems can detect and respond to an offline node.
How it works
The drill uses the Cloud Assistant plug-in ACS-ECS-NetLoss, which injects packet loss at the network layer of the target instance. Setting the packet loss ratio to rate=100 drops all inbound and outbound packets, simulating a full network outage.
Take the following precautions before starting the drill to keep the instance manageable throughout:
Configure destination IP addresses to limit the scope of packet loss to the intended targets.
Use Cloud Assistant for all command injection and recovery operations.
Set a timeout exit mechanism so the drill ends automatically if not stopped manually.
Keep the ECS console open to restart instances if needed.
Prerequisites
Before you begin, make sure you have:
An ECS instance with Cloud Assistant installed and running
Permission to run Cloud Assistant commands on the target instance
A monitoring or alerting system to observe the instance behavior during the drill (recommended)
Run a network interruption drill
Step 1: Verify the baseline
Before triggering the drill, confirm that the instance has normal network connectivity. Run a connectivity check from within the instance — for example, ping a known reachable endpoint:
ping -c 4 <reachable-endpoint>Confirm 0% packet loss. Record this result as your baseline to compare against after the drill starts.
Step 2: Start the drill
Run the ACS-ECS-NetLoss plug-in through Cloud Assistant with rate=100 to simulate a full network outage.
For the full parameter reference and step-by-step instructions, see Network packet loss drill. The network interruption drill uses the same plug-in with the packet loss ratio set to the maximum value.
Step 3: Verify the drill is active
After the drill starts, confirm that network connectivity is disrupted:
Run the same connectivity check used for the baseline.
Verify that packets are now dropped (100% packet loss).
Check that your monitoring or alerting system has detected the outage and triggered the expected alerts.
If your system relies on health checks or heartbeat signals, confirm they are reporting the node as unavailable.
Step 4: Stop the drill and recover
Use Cloud Assistant to send the recovery command to the instance to restore normal connectivity.
After recovery, run the baseline connectivity check again to confirm 0% packet loss and verify that your monitoring system reflects the instance as back online.
What's next
Network packet loss drill — run a partial packet loss drill to test degraded network conditions