Description

This article provides information about the self ping mechanism that is used  to detect the fabric single link failure in a MX-series router.

Symptoms

Information about the self ping mechanism that is used to detect the fabric single link failure in a MX-series router.

Solution

 

PFE and Switch Fabric overview :

  • Each PFE and FPC is connected via a HSL2 link.
  • The HSL2 fabric path between a source and destination PFE consists of two segments - source-fabric and fabric-destination.
  • A request-grant mechanism is used to implement the flow control.
  • Each PFE that wants to send a packet to a destination PFE sends a request; only when the request is granted, is it allowed to send the packet to the destination PFE.

MX960 with SCB and DPC :

alt

MX960 with SCB and MPC :

alt

MX960 with SCB and DPC/MPC :

alt

Grant bypass Mechanism in MX-series router (local grant) :

  • MX-series routers use the grant bypass mechanism.
  • In MX routers, the fabric chip intercepts requests from ingress the PFE and provides grants. Grants from the egress PFE are discarded by the fabric chip.

    alt



Single Link failure and Fabric self ping :

  • Single link failure means that the HSL2 link failure is in the [source PFE-fabric] or [fabric-destination] .
  • Generally, the destination error is detected by the source PFE by using the destination timeout, if the source PFE does not receive the grant for the request within a certain period of time.
  • As MX uses the grant bypass mechanism, regular destination timeout does not provide the information about the link failure on the fabric-destination link.
  • To detect failures on the fabric-destination path, each PFE in the chassis sends fabric traffic that is destined to itself and monitors the destination timeout for self-destination.

Self ping failure detection :

  • The destination timeout mechanism is used for self ping failure detection.
  • For ichip, if the destination timeout is seen in two consecutive intervals of 500 ms, the timeout is declared.
  • For mqchip, if the destination timeout is seen for any port group for the first 500 ms interval, self ping failure is declared.

Action for self ping failure :

  • PFE informs RE about the single link failure
  • RE asks all PFEs in the chassis to disable the bad destination/plane.
  • The corrective action will be different, depending on the type of hardware (Ichip, MQchip, and so on). Contact to JTAC for further assistance.

Related Information