Deploy with DVR mode for network, while shutdown controller for a while, VM with floating IP can't be accessed

Bug #1651632 reported by ShenBin on 2016-12-21
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Fuel for OpenStack
Undecided
Unassigned

Bug Description

When I deployed OpenStack with Fuel 8.0, choose DVR as network HA. If I shutdown controller node, the VMs with floating ip can't be accessed after about 10 minutes. But I learned about the VM with floating ip could be accessed when deploy with DVR mode.
This is my env:
3 nodes: 1 controller node (node-11) and 2 compute nodes (node-12, node-13);
network HA: DVR
4 VMs:
VM1 with floating ip 192.168.41.41 on node-13
VM2 with floating ip 192.168.41.42 on node-12
VM3 without floating ip on node-13
VM4 without floating ip on node-12

node-12 fip network namespace with IP 192.168.41.43
node-13 fip network namespace with IP 192.168.41.44

shutdown controller node (node-11)

ping 192.168.41.41
ping 192.168.41.42
ping 192.168.41.43
ping 192.168.41.44

192.168.41.43 and 192.168.41.44 can always accessed;
192.168.41.41 and 192.168.41.42 can't be accessed after about 10 minutes.

So why VM with floating ip can't be accessed with DVR mode if controller node shutdown?

Oleg Bondarev (obondarev) wrote :

Hi, most probably the reason is that instances are unable to renew dhcp lease (because dhcp agent is on controller node) and hence loose their private IP addresses. Please try increasing dhcp_lease_time parameter in neutron.conf

Changed in fuel:
status: New → Incomplete
Oleg Bondarev (obondarev) wrote :

Sorry, I meant dhcp_lease_duration. It can also be set to -1 for infinite

ShenBin (shenbinchina) wrote :

Thanks Oleg Bondarev , I set dhcp_lease_duration to -1 and it works!

Changed in fuel:
status: Incomplete → Fix Committed
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers