Description

This article provides an explanation for misleading Multicast statistics when the show multicast route command is executed quickly in succession.

Symptoms

Constant and continuous traffic is egressing out for all Multicast groups. When the show multicast route extensive command is executed quickly in succession at an interval of 1-2 seconds, the kBps and pps statistics report the incorrect values.

Example:

root@QFX5100> show multicast route instance Mcast-Group1 group 239.50.0.1 extensive | match stat    
    Statistics: 3223 kBps, 3223 pps , 549101 packets
    Route state: Active
    Forwarding state: Forwarding

root@QFX5100> show multicast route instance Mcast-Group1 group 239.50.0.1 extensive | match stat    
    Statistics: 0 kBps, 0 pps , 549101 packets
    Route state: Active
    Forwarding state: Forwarding

root@QFX5100> show multicast route instance Mcast-Group1 group 239.50.0.1 extensive | match stat    
    Statistics: 2877 kBps, 2877 pps , 553037 packets
    Route state: Active
    Forwarding state: Forwarding

root@QFX5100> show multicast route instance Mcast-Group1 group 239.50.0.1 extensive | match stat    
    Statistics: 0 kBps, 0 pps , 553037 packets
    Route state: Active
    Forwarding state: Forwarding

Solution

This is an expected behavior on the QFX5100. In the current QFX5100 design, the statistics are fetched from hardware by the PFE and updated to the kernel. The kernel calculates the pps and kbps based on the timestamp, and this goes out of sync between kernel and PFE if the show command is given continuously.

Allow 10-15 seconds between the execution of the show commands.