SP2000 Link Aggregation Monitoring

In this section:

Note

The following section is for the SP2000 Platforms only.

Link Aggregation (LAG) Monitoring for the SP2000 is a function of the Ethernet bonding driver that detects link faults, and is important in ensuring that the redundant network connection can be sustained in the event of a component failure. The SP2000 cards need to use LAG Monitoring because the Ethernet ports on the AMCs are not directly connected to the backplane chassis of the ATCA 13U or 3U chassis. The AMCs connect to a Ethernet Switch on the carrier blade, and it is this switch that has ports directly connected to the chassis backplane. This means that the AMC interfaces cannot directly see any loss of the Ethernet carrier signal on the backplane, and another mechanism (LAG Monitoring) is needed to allow the AMC interfaces to detect backplane link faults.

LAG Monitoring works by periodically validating the Ethernet communication path for each link from the physical network ports on an SP2000 card to the next actively managed network node, which may be a port on an SCX switch card on a 13U deployment, or a next-hop router in a 3U deployment.

The validation is done by generating an Address Resolution Protocol (ARP) packet to the SCX or next-hop router and waiting for a response. If ARP responses are being lost, the related link is flagged as being in a failed state. Any subsequent Ethernet traffic is redirected to a peer link until the validation succeeds.

To avoid being blocked by L2 routing rules in the related 3U network, the backup slave in the bond makes use of a unique MAC address for ARP validation. In addition, the LAG Monitoring logic is expanded to treat reception of an ARP query packet from the standby slave by the active slave as indication that the backup is working, and that it should not fail validation.

The LAG Monitoring statistics provide a view of historical timing of ARP reply packets from the next-hop routers, and if slow (or lost) ARP reply packets are occurring frequently which may be an indication of network congestion or some other incorrect behavior. The LAG Monitoring statistics are reported by the bonding driver in the /proc/net/ bonding/bond0 file. This file is not easily deciphered, and typically only Ribbon Global Product Support will examine the contents for troubleshooting purposes.

Note

It is recommended that the LAG Monitoring timeout default values presented in the Menu UI are not changed, but if you has a more complex network, there is the possibility of slow ARP replies that could result in unnecessary bonding failovers or loss of redundancy (that is, if the standby slave is unavailable due to an ARP validation failure). Any changes to the timing of the LAG Monitor logic requires an associated change to the Transparent Inter-Process Communication (TIPC) link threshold setting.

To enable LAG Monitoring, the bonding driver needs to know the datadefault IP addresses of the SCX cards if on a 13U chassis, or the basedefault gateway addresses of the next-hop routers (typically ERS-8600 units) if the system is 3U.

Configuring LAG Monitoring includes the following procedures.

To query the datedefault IP address on the 13U

Click to read more...

To configure the LAG Monitoring timeout

Click to read more...

Log onto the either of the SP2000 Management CPUs.
At the prompt, enter startmenu, and press ENTER.
Enter 2, and press ENTER.
Enter 2, and press ENTER.
Enter 7, and press ENTER.
Enter 2, and press ENTER.
Enter the number for the required Management CPU (in this example, 1).
To begin configuring AMC121 Management CPU in slot14, and the associated AMC308 I/O cards on the same carrier as slot14, enter 1, and press ENTER.
Enter 5 to query any existing LAG Monitoring IP addresses, and press ENTER.
Do one of the following:
- If no previous LAG Monitoring IP configuration exists as shown in the following screen capture and the output of the query command does not show any IP addresses, do the following:
  1. Press ENTER.
  2. Enter 7, and press ENTER. Go to step step12.
- If the output of the query command shows that LAG Monitoring IP addresses already exist as shown in the following screen capture, these IP addresses must be cleared before you can continue this procedure.
  In this scenario, one of the following (or both) outputs may appear on your screen. Go to step a.
  1. Press ENTER.
  2. Enter 7, and press ENTER to clear the existing LAG Monitoring IP addresses.
    
    Tip
    
    If the preceding Warning continues to be displayed after you performed this task, contact Customer Support (refer to About the Documentation Library for contact details).
Enter 6, follow the prompts to enter the IP Addresses, and press ENTER.
After entering the IP Addresses, the current LAG Monitoring timeout is displayed, along with a prompt asking you whether you want to change the value for this timeout.
To reconfigure the LAG Monitoring timeout, enter y, and press ENTER.
Re-enter the LAG Monitoring timeout value (even if it is the same) to reset the internal LAG Monitoring timeout configuration.

Tip

In most configurations, the default value should be used.
Press ENTER.
Repeat this procedure for the Management CPU in slot24.

Tip

Make sure to use the appropriate IP Addresses if the system is in a Geographic Redundancy configuration.
Reboot all system cards for the changes to take effect.