AAP/FIP: incorrect stitched MAC for FIP breaks L2 connectivity
Bug #1615130 reported by
amit surana
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
Juniper Openstack | Status tracked in Trunk | |||||
R3.0 |
Fix Committed
|
Medium
|
Naveen N | |||
R3.1 |
Fix Committed
|
Medium
|
Naveen N | |||
Trunk |
Fix Committed
|
Medium
|
Naveen N |
Bug Description
Consider a case where a FIP from a public VN has been associated with an AAP (which in-turn front ends a vrrp group that has 2 VMs). This is done by setting the 'floating_
In the public VN (from which FIP is taken), the L3 vrf has a stitched MAC programmed for the FIP, which is set to the MAC of that VM to which the FIP was assigned last. Now, if a VM in the public VN tries of ping the FIP, its ARP request will be proxies by vRouter and replied to with the stitched MAC. As such, the echo request will always be bridged to the 2nd VM -- irrespective of which one is active in the vrrp group.
summary: |
- AAP/FIP: stitched MAC for FIP breaks L2 connectivity + AAP/FIP: incorrect stitched MAC for FIP breaks L2 connectivity |
To post a comment you must log in.
When floating-ip is associated with 2 VM, there will be 2 EVPN routes exported since mac of each VM is different. If tracking IP is set on FIP, preference of FIP would be tracked based on the tracking-ip route. MAC stitching of VM also should take into account preference and change the stitched MAC upon preference change