Revision history [back]

click to hide/show revision 1
initial version

VMs do not finish booting on first node in two node RDO icehouse installation

I've installed a standard RDO (icehouse) allinone setup. After that I've added additional node.

I had the following issue: https://ask.openstack.org/en/question/30090/network-is-slow-on-compute-node-of-two-node-icehouse-neutron-ovsvlan-centos-65-fedora-20-cluster/; which was resolved as suggested in the link. BUT, for some reason I cannot now launch VMs on the first node (which is the all in one). I don't remember any configuration change.

Quite strange really that running VMs on the second (compute) node is fine, but the first one doesn't work. For example, the cirros image doesn't get ssh key injected.

Any tips and suggestions are welcome.

VMs do not finish booting on first node in two node RDO icehouse installation

I've installed a standard RDO (icehouse) allinone setup. After that I've added additional compute node.

I had the following issue: https://ask.openstack.org/en/question/30090/network-is-slow-on-compute-node-of-two-node-icehouse-neutron-ovsvlan-centos-65-fedora-20-cluster/; which was resolved as suggested in the link. BUT, for some reason I cannot now launch VMs on the first node (which is the all in one). one).

I don't remember any configuration change.

Quite strange really changed, but it seems that running VMs launching on the second (compute) first node is fine, but the first one doesn't work. For have an issue with networking or metadata (they can't reach 169.254.169.254), also, for example, the cirros image doesn't images do get IPs and network seems to work in them, but they don't get ssh key keys injected.

Any tips and suggestions are welcome.

VMs do not finish booting on first node in two node RDO icehouse installation

I've installed a standard RDO (icehouse) allinone setup. After that I've added additional compute node.

I had the following issue: https://ask.openstack.org/en/question/30090/network-is-slow-on-compute-node-of-two-node-icehouse-neutron-ovsvlan-centos-65-fedora-20-cluster/; which was resolved as suggested in the link. BUT, for some reason I cannot now launch VMs on the first node (which is the all in one).

I don't remember any configuration changed, but it seems that VMs launching on the first node have an issue with networking or metadata (they can't reach 169.254.169.254), also, for example, the images do get IPs and network seems to work in them, but they don't get ssh keys injected.

Any tips and suggestions are welcome.

Update:

Just checked the following:

  1. Starting a VM (cirros) on the second node happens fast, DHCP request is handled once and fast, metadata contacted fine.
  2. Starting a VM (cirros) on the first node (where controller and network are running) is happening slowly, DHCP request handled twice, metadata is not reached.

So I imagine the issue is network-related config on the first node. But what can it be?

Thanks.