Managed OpenStack
Latest
Ingress Traffic Being Flooded Unexpectedly for Instances on Provider Network
Problem
Virtual machines (VMs) residing on a provider, VLAN-based network are receiving traffic not directed to them – resulting in increased overhead and performance degradation.
Example:
root@odr1:~# ip addr...2: ens3: [BROADCAST,MULTICAST,UP,LOWER_UP] mtu 1500 qdisc pfifo_fast state UP group default qlen 1000link/ether fa:16:3e:53:88:d1 brd ff:ff:ff:ff:ff:ffinet 10.200.3.74/16 brd 10.200.255.255 scope global ens3valid_lft forever preferred_lft foreverroot@odr1:~# tcpdump -eni ens3 'ip and not host 10.200.3.74'tcpdump: verbose output suppressed, use -v or -vv for full protocol decodelistening on ens3, link-type EN10MB (Ethernet), capture size 262144 bytes15:47:15.433385 fa:16:3e:b9:3a:e7 > fa:16:3e:44:37:2e, ethertype IPv4 (0x0800), length 46: 92.xxx.xxx.x > 10.200.4.217: ICMP echo request, id 39603, seq 1, length 1215:47:16.392894 fa:16:3e:b9:3a:e7 > fa:16:3e:44:37:2e, ethertype IPv4 (0x0800), length 46: 167.xxx.xx.x > 10.200.4.217: ICMP echo request, id 14148, seq 1, length 12...15:47:18.871963 fa:16:3e:b9:3a:e7 > fa:16:3e:fe:e0:55, ethertype IPv4 (0x0800), length 46: 92.xxx.xxx.x > 10.200.1.221: ICMP echo request, id 19140, seq 1, length 1215:47:20.637640 fa:16:3e:b9:3a:e7 > fa:16:3e:fe:e0:55, ethertype IPv4 (0x0800), length 46: 92.xxx.xxx.x > 10.200.1.221: ICMP echo request, id 55584, seq 1, length 12
Environment
Platform9 Managed OpenStack - v4.0 – v4.3
Neutron
OVS
VLAN – Provider Network
Cause
Resolution
The related bug has been addressed upstream and Platform9 has backported to all affected PMO versions.
Last updated on
Was this page helpful?
Frequently Asked Questions
Solution
How To
Internal Only
Templates
Powered by
