Description

This article describes the issue of the RETH interface being up for 10 or more seconds, even after the child interfaces are physically down.

Symptoms

  • In some high-availability SRX deployments, the child interfaces of a RETH interface are connected to the same switch.
  • In such a scenario, if the switch crashes or loses power, both the child interfaces go down at once.
  • However, contrary to intuitive assumption, the parent RETH interface will stay up for 10 or more seconds.

If a dynamic routing protocol is running on a RETH interface, no re-convergence will take place until the RETH interface is marked as down . This could result in traffic getting black-holed for 11-15 seconds, until the routing protocol re-converges and traffic takes an alternate path.

Solution


  • The behavior of the RETH interface is as designed. Juniper does not recommend connecting all child interfaces to the same switch, as this creates a single point of failure.
  • So, this kind of topology is not considered common behavior.
  • In order to avoid such a situation, connect the child interfaces to different switches.