DHCP in kolla-ansible / VMWare / DVS setting

asked 2019-05-14 09:45:49 -0500

updated 2019-05-15 09:36:16 -0500

Update 2019-05-15:

Some progress I believe. However, I still don't see dhcp requests from my test vm arriving at the OpenStack network nodes. What I do not quite understand: When I create a flat network in OpenStack a new VMWare port group on the VMWare distributed switch is created and the vm attaches to it. However:

  • First, the vm does not attached to any other OpenStack control network / VMWare port group (at least I don't see an appropriate nic)
  • Second, apparently none of the OpenStack network nodes attaches to the created flat network / VMWare port group.

Hence, I don't understand how vm dhcp requests are supposed to reach the network nodes in the first place. Do I have to create an appropriate nic on the vm or on the network nodes by myself?

Current status: I did notice that on my network nodes some deployment issues regarding the network interfaces must have happended since the ovs db was empty. A reboot of the network notes and re-deployment of OpenStack did seem to help.

Current ovs db:

root@network01:~# ovs-vsctl show
7465c5b6-6e70-43c7-843d-522214340362
    Bridge br-dvs
        Port br-dvs
            Interface br-dvs
                type: internal
        Port "tapde20750b-c2"
            tag: 4095
            Interface "tapde20750b-c2"
                type: internal
        Port "ens192"
            Interface "ens192"

Then, I saw that the neutron-dhcp-agent tried to apply some iptables rules. Initially that failed but I could fix it by loading the ip6-tables kernel module manually on the network nodes. However, when listing all iptables rules I cannot find any neutron-specific rules - Maybe this is an issue right now?

iptables:

root@network01:~# sudo iptables --list
Chain INPUT (policy ACCEPT)
target     prot opt source               destination

Chain FORWARD (policy ACCEPT)
target     prot opt source               destination
DOCKER-USER  all  --  anywhere             anywhere
DOCKER-ISOLATION-STAGE-1  all  --  anywhere             anywhere
ACCEPT     all  --  anywhere             anywhere             ctstate RELATED,ESTABLISHED
DOCKER     all  --  anywhere             anywhere
ACCEPT     all  --  anywhere             anywhere
ACCEPT     all  --  anywhere             anywhere

Chain OUTPUT (policy ACCEPT)
target     prot opt source               destination

Chain DOCKER (1 references)
target     prot opt source               destination

Chain DOCKER-ISOLATION-STAGE-1 (1 references)
target     prot opt source               destination
DOCKER-ISOLATION-STAGE-2  all  --  anywhere             anywhere
RETURN     all  --  anywhere             anywhere

Chain DOCKER-ISOLATION-STAGE-2 (1 references)
target     prot opt source               destination
DROP       all  --  anywhere             anywhere
RETURN     all  --  anywhere             anywhere

Chain DOCKER-USER (1 references)
target     prot opt source               destination
RETURN     all  --  anywhere             anywhere

No dhcp request at network nodes:

root@network01:~# cat /var/log/syslog | grep DHCP
root@network01:~#

Dear community,

currently, we are in the process of deploying OpenStack Stein for evaluation purposes. Our target platform (hypervisor) is a VMWare environment in version 6.0.0 (only targeting provider networks without NSX for now). OpenStack nodes are deployed as VMWare VMs in one cluster (os: Ubuntu), tenant VMs should be spawned in a second cluster. We deploy OpenStack via kolla-ansible (8.0.0.0rc1) and followed a default multinode inventory with three control and two network nodes.

Our main problem is that VMs (cirros) can be spawned in a provider network created via OpenStack (dhcp enabled), however, they don't get an ip configuration via dhcp. Now, we are unsure how to ... (more)

edit retag flag offensive close merge delete