Ask Your Question

sasok's profile - activity

2014-09-11 03:04:21 -0500 received badge  Good Question (source)
2014-04-21 19:22:50 -0500 received badge  Famous Question (source)
2013-11-25 14:36:44 -0500 received badge  Notable Question (source)
2013-10-24 13:01:32 -0500 received badge  Famous Question (source)
2013-10-23 23:26:12 -0500 received badge  Famous Question (source)
2013-09-21 03:52:43 -0500 received badge  Nice Question (source)
2013-09-21 03:51:14 -0500 received badge  Notable Question (source)
2013-09-10 17:47:15 -0500 received badge  Popular Question (source)
2013-09-09 08:38:01 -0500 received badge  Notable Question (source)
2013-09-08 07:57:12 -0500 received badge  Enthusiast
2013-09-07 08:12:40 -0500 asked a question Devstack single node provider network possible?

I am having problems configuring vlan provider network on a single node devstack install. i would like to know if provider network are even possible on a single node install, because VMs can communicate directly over br-int integration bridge and physical connection isn't even used. I have configured a separate br-pr bridge and added eth2 interface. I created a vlan provider network on a configured physnetpr physical connection that uses br-pr bridge. I have connected a laptop to eth2 interface and configured vlan 100 on its ethernet interface. I have a VM running on this provider network, but communication between VM and laptop does not seem to work. I should also mention that I have disabled security groups by setting OVS firewall to NoopFirewallDriver. When using wireshark to look at the packets, even ARP packets do not get to their destination. ARP packets from the laptop seem to come to int-br-pr interface and ARP packets from the VM seem to come phy-br-pr interface. Is what I am trying to do even possible?

2013-09-04 09:41:52 -0500 received badge  Popular Question (source)
2013-09-04 08:24:17 -0500 commented answer Flat provider network without dhcp not working

Thanks. Do you know if this mechanism can be disabled in any way for the port?

2013-09-03 16:54:06 -0500 asked a question Flat provider network without dhcp not working

I am trying to configure flat provider network on a single-node devstack install. Normal networking and floating iPs all working as they should. I have added another physical interface (eth2) to be used for a flat provider network. I have created a flat provider network attached to eth2. I had to also create a subnet on the provider network, otherwise instance creation of an instance connected to the provider network was not possible. CIDR for the subnet is 192.168.10/24. I have disabled DHCP for this subnet. On the instance attached to this provider network, the IP on the interface is configured statically to be 192.168.10.10. I have another machine connected physically to eth2 with the IP 192.168.10.12. When I try to ping from 192.168.10.12 to 192.168.10.10, the replies never come back. ARP seems to be working OK. When using tshark to inspect various interfaces, replies can also be seen on the tap interface and qbr interfaces, so the instance seems to be replying. But on the veth pair attached to qbr the replies are no longer present.

When I inspect the instance with nova show <instance_id> I can see nova showing IP 192.168.10.2. Subnet has DHCP disabled, but still this IP is shown. When I set the IP address of the instance interface to this IP, everything seems to be working fine. I can ping from the instance to the outside machine and vice versa. So the connectivity only seems to be assured for the IP that nova has chosen.</instance_id>

I would like to ask if this behaviour is intended or am I misconfiguring something. Would configuring the provider network as type vlan help?

2013-06-07 17:36:00 -0500 received badge  Popular Question (source)
2013-05-06 16:07:00 -0500 asked a question Is there a simple way to purge tenant/user together with all instances, networks etc. ?

Does a simple command exist in openstack that would remove all tenant and/or user infrastructure together with instances, networks floating ips etc.? This would effectively purge tenant together with all traces of him from OpenStack. I have tried keystone tenant-delete and user-delete, but this does not seem to delete instances etc. The instances seem to remain running after this.

2013-04-19 02:26:04 -0500 received badge  Student (source)
2013-04-16 08:16:42 -0500 received badge  Famous Question (source)
2013-04-10 01:49:17 -0500 received badge  Notable Question (source)
2013-04-04 14:59:10 -0500 received badge  Supporter (source)
2013-04-04 14:59:04 -0500 commented answer How many network interfaces needed on controller/network node?

Thanks Anne. I failed to specify in my question above, that I am using Quantum for networking. Operations guide only focuses on nova-network, if I am not mistaken. Nevertheless, I will check the relevant sections.

2013-04-04 14:53:58 -0500 received badge  Popular Question (source)
2013-04-03 19:00:17 -0500 received badge  Editor (source)
2013-04-02 15:11:46 -0500 asked a question How many network interfaces needed on controller/network node?

I am setting up an OpenStack Folsom setup on two nodes. I am using Quantum for networking. First node controller/network and the second node compute. I would like the VMs to have access to external network with floating ips etc. The question is how do I configure the controller/network node and how many network interfaces I need. If I understand correctly I need one interface for management and one for VM data network (from compute node). Here comes the tricky part. Do I need two more interfaces - one for VM external net and one for OpenStack API/dashboard access or can these two functions be configured on the same network interface? If these two things can be configured on the same interface, how do I do this?

Thanks for your help.