Agent should not connect to contrail-dns when it doesnt have end of RIB
Bug #1605341 reported by
Hari Prasad Killi
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
Juniper Openstack | Status tracked in Trunk | |||||
R2.21.x |
Fix Committed
|
Medium
|
Nipa | |||
R3.0 |
Fix Committed
|
Medium
|
Nipa | |||
R3.1 |
Fix Committed
|
Medium
|
Nipa | |||
R3.2 |
Fix Committed
|
Medium
|
Nipa | |||
R4.0 |
Fix Committed
|
Medium
|
Nipa | |||
Trunk |
Fix Committed
|
Medium
|
Nipa |
Bug Description
When contrail-dns doesnt have end of RIB, connections from agent should not be successful. If the connection is closed and agent sends a re-subscribe to discovery after a few retries, agent may move to a contrail-dns which can provide service.
To post a comment you must log in.
On master image, Agent has a list of all DNS publishers in the system as agent subscribes with instances=0. When Agent detects a TCP connection DOWN (as control-node was being upgraded), agent after a few retries will try other DNS Servers in the list. Note at this point the upgraded DNS Servers which may be in the publisher list( as ttl did not expire), may be able to accept TCP connections. These new TCP connections need to be closed if end-of-rib is not computed.