NETVIRT-1520: AlivenessMonitor cache issue 83/78083/7
authorAnil Kumar Gujele <anilkumar.g@altencalsoftlabs.com>
Fri, 23 Nov 2018 09:44:49 +0000 (15:14 +0530)
committerSam Hague <shague@redhat.com>
Fri, 25 Jan 2019 03:22:32 +0000 (03:22 +0000)
commitb973935efe722319f1fb6b4749bc96830f125e8c
tree2d36fff99af6404c5cacd0ab656393600c505f92
parente2eaa55eb63d4e6724c88f1b6307655982deddb6
NETVIRT-1520: AlivenessMonitor cache issue

Issue:
Say VM with MIP-IP is on DPN connected to Contorller-1, when MIP is
learnt , it populates the cache alivenessCache. When VM migrates to
another DPN, which is connected to say Controller-2, the same
alivenessCache will not have data w.r.t the monitoring session
created by Controller-1.
Hence incase of cleanup , it causes issues in stopping ArpMonitoring.
The cache is simple local cache which doesnt work incase of cluster.

Fix:
Better to read InterfaceMonitorMap directly from datastore so that
monitorId can be retrieved even if MIP moves across DPNs connected to
various controllers in cluster.

Change-Id: I1e4102a7c57bff8376d3d05be8975a6c1ad74233
Signed-off-by: Anil Kumar Gujele <anilkumar.g@altencalsoftlabs.com>
vpnmanager/impl/src/main/java/org/opendaylight/netvirt/vpnmanager/iplearn/AlivenessMonitorUtils.java
vpnmanager/impl/src/main/java/org/opendaylight/netvirt/vpnmanager/iplearn/IpMonitorStopTask.java