[CLOUDSTACK-9189] rVPC ACL doesn't recover after cleaning up through the NetworkGarbageCollector

• Type: Bug
• Status: Open
• Priority: Critical
• Resolution: Unresolved
• Affects Version/s: 4.7.0
• Fix Version/s: Future
• Component/s: Virtual Router
• Security Level: Public (Anyone can view this level - this is the default.)
• Assignee: Wilder Rodrigues
• Reporter: Wilder Rodrigues
• Created: 19/Dec/15 15:50
• Updated: 13/Apr/16 02:57

Description

In order to reproduce this bug quicker is better to change the network.gc.interval and gc.wait from 600 seconds to 10 seconds via Global Settings and restart your management server.

• deploy a rVPC
• deploy VM in it
• make port forwarding (2nd ip, firewall and such)
• confirm it works
• stop the vm
• after some time (20 seconds * 3 - approximately) the network garbage collector will come and tear down the network since there are no more VMs
• all the nics will be fine and the guest nic will be gone. The routers should be on BACKUP/BACKUP
• then start the vm again
• the nics get plugged again and keepalived will decide on a new master.
• try to SSH into the VM via the public IP. It will fail

The only way to get it working afain is:

• Replace the network ACL - for example, default allow all
• Try to SSH again and it works fine
• Replace back to your original ACL
• Try to SSH again and it works fine