VM instance cannot be pinged or SSH
I have used a provider network based on the tutorial setup for Rocky Installation. I have 2 VMs fully working where I can ping them and also SSH into them using the keypair.
When I try creating my third VM, the instance gets an IP but I am unable to ping. When I keep on deleting an creating new instances, sometimes I am able to ping the new instance but still unable to ssh. I get connection refused error.
When I looked into the IPtables of the compute node I see that the new entries are not added when I create an instance. If it gets added I am able to ping the VM but still cannot SSH.
The current setup is using DHCP I assume: https://imgur.com/0zTswHZ The ports are also created: https://imgur.com/vr8ry3A instance creation : openstack server create --flavor mi.large --image ubuntuServer --nic net-id=111fd1bd-b3bb-4085-8022-d1354c7d016f --security-group default --key-name n1 n1
Could anyone please provide an idea as to what is happening? Update: I am still able to create the instance without any error. But when I checked the logs I found it had 2 errors:
- amqp server not reachable controller:5672 error. So I created a new openstack rabbitmq user, updated all the config files to use the new user and then this error disappeared.
- The second error in the compute log was:
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager [req-25b3f5d8-2831-4070-a08f-e101b6dc5a2e - - - - -] Error updating resources for node compute2.: DiskNotFound: No disk at /var/lib/nova/instances/4c9c7d22-b0c0-4342-9b02-d9956923065f/disk
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager Traceback (most recent call last):
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager File "/usr/lib/python2.7/dist-packages/nova/compute/manager.py", line 7736, in _update_available_resource_for_node
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager rt.update_available_resource(context, nodename)
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager File "/usr/lib/python2.7/dist-packages/nova/compute/resource_tracker.py", line 708, in update_available_resource
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager resources = self.driver.get_available_resource(nodename)
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager File "/usr/lib/python2.7/dist-packages/nova/virt/libvirt/driver.py", line 6540, in get_available_resource
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager disk_over_committed = self._get_disk_over_committed_size_total()
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager File "/usr/lib/python2.7/dist-packages/nova/virt/libvirt/driver.py", line 8162, in _get_disk_over_committed_size_total
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager err_ctxt.reraise = False
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager File "/usr/lib/python2.7/dist-packages/oslo_utils/excutils.py", line 220, in __exit__
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager self.force_reraise()
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager File "/usr/lib/python2.7/dist-packages/oslo_utils/excutils.py", line 196, in force_reraise
2020-06-18 12:43:22.762 1364 ERROR nova.compute.manager six.reraise(self.type_, self.value, self ...
Provider networks usually don’t serve dhcp requests, depending on how your instance is configured to obtain its IP it could be necessary to use config-drive for the boot command instead of dhcp. Can you provide more details about dhcp disabled on the provider network and the instance configuration?
The current setup is using DHCP I assume: https://imgur.com/0zTswHZ The ports are also created: https://imgur.com/vr8ry3A instance creation : openstack server create --flavor mi.large --image ubuntuServer --nic net-id=111fd1bd-b3bb-4085-8022-d1354c7d016f --security-group default --key-name n1 n1
Do you have an external dhcp server for the provider network? It could be a different issue but I would try it this way: in our company we can only use config-drive for provider networks. So DHCP is disabled for that subnet and config-drive applies the IP address to the VM that openstack assigned.
No I don't have any external dhcp server for provider network. My setup has 1 controller and 2 compute nodes. RabbitMQ is installed on controller.