Revision history [back]

Since your disk space for volumes is lower than the minimum volume size of 1GB, you can’t create any more volumes. If you launch instances with a volume as root disk (that’s the default if you use the GUI), this must fail.

Suggestions: Provide more space for volumes, or launch instances without volumes.

If instances without volumes can’t be launched, the problem is elsewhere. In this case, post error details and log messages.

Since your disk space for volumes is lower than the minimum volume size of 1GB, you can’t create any more volumes. If you launch instances with a volume as root disk (that’s the default if you use the GUI), this must fail.

Suggestions: Provide more space for volumes, or launch instances without volumes.

If instances without volumes can’t be launched, the problem is elsewhere. In this case, post error details and log messages.

EDIT: Instances can use so-called ephemeral storage, which disappears when the instance is deleted. In this case, their root disk is located on the compute node's filesystem, usually in the form of a file such as a qcow2 file. The size of an instance's ephemeral root disk is defined by the flavour.

You say that launching an instance with ephemeral storage fails. This means that something else is amiss; perhaps you simply don't have enough disk space or the compute node is lacking other resources like RAM.

Alternatively, an instance can use permanent storage in the form of a Cinder volume.

As I said before, when you launch an instance from Horizon, by default permanent storage is selected. You can change that with a switch in the instance launch dialog. On the other hand, when you use the CLI, the default is ephemeral storage; see https://docs.openstack.org/nova/latest/user/launch-instance-from-image.html.

Of course, Cinder can be extended. You can, for example, add a backend. You seem to be using DevStack or PackStack to deploy your cloud; these deployment tools create a small LVM volume group on a loopback disk as their sole Cinder backend. I suppose that they have option for sizing this disk; personally, I have never tried increasing the size on a runnning DevStack or PackStack, but it should be possible by:

  1. increasing the size of the disk file that underlies your loopback disk. The fallocate command may be able to do that.
  2. adapting the size of the physical volume with the pvresize command.

It's probably easier, and teaches you more about OpenStack rather than Linux administration, to add a backend. This should be documented at https://docs.openstack.org/cinder/latest/admin/blockstorage-multi-backend.html.

Since your disk space for volumes is lower than the minimum volume size of 1GB, you can’t create any more volumes. If you launch instances with a volume as root disk (that’s the default if you use the GUI), this must fail.

Suggestions: Provide more space for volumes, or launch instances without volumes.

If instances without volumes can’t be launched, the problem is elsewhere. In this case, post error details and log messages.

EDIT: Instances can use so-called ephemeral storage, which disappears when the instance is deleted. In this case, their root disk is located on the compute node's filesystem, usually in the form of a file such as a qcow2 file. The size of an instance's ephemeral root disk is defined by the flavour.

You say that launching an instance with ephemeral storage fails. This means that something else is amiss; perhaps you simply don't have enough disk space or the compute node is lacking other resources like RAM.

Alternatively, an instance can use permanent storage in the form of a Cinder volume.

As I said before, when you launch an instance from Horizon, by default permanent storage is selected. You can change that with a switch in the instance launch dialog. On the other hand, when you use the CLI, the default is ephemeral storage; see https://docs.openstack.org/nova/latest/user/launch-instance-from-image.html.

Of course, Cinder can be extended. You can, for example, add a backend. You seem to be using DevStack or PackStack to deploy your cloud; these deployment tools create a small LVM volume group on a loopback disk as their sole Cinder backend. I suppose that they have option options for sizing this disk; personally, I have never tried increasing the size on a runnning DevStack or PackStack, but it should be possible by:

  1. increasing the size of the disk file that underlies your loopback disk. The fallocate command may be able to do that.
  2. adapting the size of the physical volume with the pvresize command.

It's probably easier, and teaches you more about OpenStack rather than Linux administration, to add a backend. This should be documented at https://docs.openstack.org/cinder/latest/admin/blockstorage-multi-backend.html.

Since your disk space for volumes is lower than the minimum volume size of 1GB, you can’t create any more volumes. If you launch instances with a volume as root disk (that’s the default if you use the GUI), this must fail.

Suggestions: Provide more space for volumes, or launch instances without volumes.

If instances without volumes can’t be launched, the problem is elsewhere. In this case, post error details and log messages.

EDIT: Instances can use so-called ephemeral storage, which disappears when the instance is deleted. In this case, their root disk is located on the compute node's filesystem, usually in the form of a file such as a qcow2 file. The size of an instance's ephemeral root disk is defined by the flavour.

You say that launching an instance with ephemeral storage fails. This means that something else is amiss; perhaps you simply don't have enough disk space or the compute node is lacking other resources like RAM.

Alternatively, an instance can use permanent storage in the form of a Cinder volume.

As I said before, when you launch an instance from Horizon, by default permanent storage is selected. You can change that with a switch in the instance launch dialog. On the other hand, when you use the CLI, the default is ephemeral storage; see https://docs.openstack.org/nova/latest/user/launch-instance-from-image.html.

Of course, Cinder can be extended. You can, for example, add a backend. You seem to be using DevStack or PackStack to deploy your cloud; these deployment tools create a small LVM volume group on a loopback disk as their sole Cinder backend. I suppose that they have options for sizing this disk; personally, I have never tried increasing the size on a runnning DevStack or PackStack, but it should be possible by:

  1. increasing the size of the disk file that underlies your loopback disk. The fallocate command may be able to do that.
  2. adapting the size of the physical volume with the pvresize command.

It's probably easier, and teaches you more about OpenStack rather than Linux administration, to add a backend. This should be documented at https://docs.openstack.org/cinder/latest/admin/blockstorage-multi-backend.html.

Or find out with what option you can specify the Cinder backend size and redeploy your cloud from scratch.