Reset Search



LAG configured with LACP does not failover as expected

« Go Back


TitleLAG configured with LACP does not failover as expected
  • When doing failover tests on a share group (LAG), pings across the LAG drop when one of the links goes down even though the redundant link is still active.
  • Long LACP failover times
  • LACP failover problems when media converters are placed in-line
  • EXOS
  • Summit
  • BlackDiamond
  • LACP timeout

There are two different scenarios which can cause a LACP link failover:

  1. Link down of one LAG member port- EXOS will not wait for LACP timeout and immediately removes that port from LAG membership.  Traffic gets immediately switched to next available active port in LAG.
  2. Link stays up but no LACP PDU received from peer switch- EXOS will take minimum 3 seconds (if set to short) for LACP timeout to happen and then remove that port from LAG membership. In this case, traffic switchover will take 3 seconds.
In a scenario where an actual link down occurs, failover should occur immediately upon detection of the link down.  If the link is not going down (i.e. due to some media converter in between) the LACP timeout timer will be the determining factor in link failover.  

The default timeout period for LACP in EXOS is 90 seconds.  LACP timeout can be configured as either long (90 seconds) or short (3 seconds).  If the default timeout is too long, it can be reduced by typing the command "configure sharing <port> lacp timeout short". The down link will now be removed from the aggregator after 3 seconds.

Additional notes
  • The switch will send LACP PDUs, with timeout flag set, every second if using short timers 
  • Please note that if you disable the sharing and then re-enable , you will have to re-enter the command  "configure sharing <port> lacp timeout short" since this will revert to defaults again.



Was this article helpful?



Please tell us how we can make this article more useful.

Characters Remaining: 255