process-restart

Usage Guidelines

Note

The process restart capability should not be enabled when BGP EVPN is configured. For more information, see the Extreme SLX-OS Layer 3 Routing Configuration Guide.

The process restart capability for BGP, IS-IS, MPLS, OSPFv2 and OSPFv3 is a fault containment mechanism that ensures that process-level failures do not cause system-level failures.

All of these processes are COLD restartable processes, meaning when there is a fault inside one of these processes, and it crashes, the system does not undergo a failover. Instead, the process is restarted on the same ACTIVE MM. All the other modules and processes that interact with this particular process are made aware of the process restart, and they adjust accordingly.

All MPLS-based services, such as IPoMPLS, VLL, and VPLS, are disrupted for the duration of MPLS process restart. After the MPLS process is restarted, the control protocols (LDP and RSVP) re-signal the tunnels and cross-connects and all the dependent MPLS applications to resume service. For routing protocols such as BGP, IS-IS and OSPF, all the existing adjacencies are brought down, and are reestablished after the process comes up again. Some traffic loss can be expected until the adjacency comes up again.

If more than one process restart happens within a minute, the entire board reboots, resulting in device reload.

The benefits of the process restart feature can be seen in the following cases:

• The unplanned outage of the routing protocol process can be handled without impacting the whole system or other dependent services.
• The system returns to a stable state quicker than if nonstop routing (NSR) or graceful restart (GR) is configured on the device.

The following table lists the precedence of process restart over GR or NSR.