Make sure that the bridges are being created (1 bridge per vm) they should be 
named close to the vm tap device name.  Then make sure that you have bridge 
nf-call-* files enabled:

http://wiki.libvirt.org/page/Net.bridge.bridge-nf-call_and_sysctl.conf

Under hybrid mode what happens is a linux bridge (not an ovs bridge (brctl)) is 
created per vm.  The vm's tap device is plugged into this bridge.  A veth is 
created that spans from the vm's linux bridge to br-int and is plugged at both 
ends.  This is done because older versions of OVS did not have support (or 
efficient support) for doing firewalling.  The problem is that in the kernel, 
packets traversing the Openvswitch code paths are unable to be hooked into by 
netfilter.  So the linux bridge is created solely to allow the VM traffic to 
pass through a netfilter hookable location, so security groups work.

You need at a minimum to make sure /proc/sys/net/bridge/bridge-nf-call-iptables 
is set to 1.  If its not then when you look at the iptables rules that are 
created – you will see that none of the security group chains are seeing 
traffic.
___________________________________________________________________
Kris Lindgren
Senior Linux Systems Engineer
GoDaddy

From: raju <[email protected]<mailto:[email protected]>>
Date: Thursday, April 21, 2016 at 5:26 PM
To: 
"[email protected]<mailto:[email protected]>"
 
<[email protected]<mailto:[email protected]>>
Subject: [Openstack-operators] Security group rules not working on instances 
kilo

Hi,

I am running into a issue where security group rules are not applying to 
instances when I create a new security group with default rules it should 
reject all incoming traffic but it is allowing everything without blocking

here is my config for nova :

security_group_api = neutron
firewall_driver = nova.virt.firewall.NoopFirewallDriver

and in ml2.con.ini

firewall_driver = 
neutron.agent.linux.iptables_firewall.OVSHybridIptablesFirewallDriver

iptables service is running on all the nodes, please suggest me if  I miss 
anything.


Thanks.
_______________________________________________
OpenStack-operators mailing list
[email protected]
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators

Reply via email to