ICMP echo requests larger than the MTU are not receiving replies when sent with the DF (Don't Fragment) bit disabled. This behavior occurs in two specific scenarios:
In both cases, no ICMP responses are received, leading to questions about whether CSA drops fragmented packets with the DF bit disabled.
Cisco Secure Access drops fragmented packets in both underlay and overlay scenarios. This behavior is documented in the Cisco Secure Access Help documentation, which explicitly states: "Fragmented packets in the underlay or overlay are dropped."
Cisco Secure Access is designed to drop fragmented packets regardless of whether they occur in the underlay or overlay network. This applies to:
ICMP packets sent from RAVPN endpoints that exceed the VPN interface MTU with DF bit cleared
ICMP packets sent from on-premise endpoints over IPsec tunnels that exceed the tunnel interface MTU with DF bit cleared
This behavior is consistent across all scenarios involving fragmented packets within the Cisco Secure Access infrastructure.
Feature request CSE-I-5739 has been created for this.
Cisco Secure Access is architected to drop fragmented packets as a security and performance design decision. This behavior is implemented to prevent potential security vulnerabilities and processing overhead associated with packet reassembly in both underlay and overlay network scenarios.
| Revision | Publish Date | Comments |
|---|---|---|
1.0 |
04-Jun-2026
|
Initial Release |