Use libvirt Storage Pools

https://blueprints.launchpad.net/nova/+spec/use-libvirt-storage-pools

Currently, the libvirt driver does not make use of libvirt’s storage pools and volumes. Using libvirt storage pools would simplify adding support for new image backends, as well as facilitating cold migrations (see follow up blueprint).

Problem description

Currently, Nova’s libvirt driver does not make any use of libvirt volumes and storage pools.

This means that, for the image backends, we have a lot of code that deals directly with various images backend formats, and we have to manually deal with a variety of different situations via various command line tools and libraries.

However, much of this functionality is already present in libvirt, in the form of libvirt storage pools, so the libvirt driver duplicates functionality already present in libvirt itself.

Use Cases

Developer: This will facilitate removing SSH from resize/migrate, as it will allow us to use virStorageVolUpload and virStorageVolDownload to migrate storage.

Developer: Second, this will simplify adding in support for pool types that are supported by libvirt but not supported by Nova (such as Sheepdog).

Developer: this will (in the long run) simplify the imagebackend code, making it easier to maintain.

Proposed change

The cache of images downloaded from Glance would be placed into a volume pool (nova-base-images-pool). This is done simply by instructing libvirt that Nova’s image cache directory (e.g. /var/lib/nova/_base) is a directory storage pool, and as such does not affect directory layout (and is thus compatible with both the legacy image backends and the new image backend proposed below).

A new image backend, LibvirtStorage, would be introduced. This would support being used in place of all of the current types (with the exeception of RBD support, which for the time being would need a subclass [1]).

If we are not using COW, the libvirt pool.createXMLFrom method could be used to appropriately copy the template image from the source pool, nova-base-images-pool, into the target image in the target pool nova-disks-pool. This works regardless of the source and destination formats (for instance, the same function calls are used to copy from raw to LVM or from qcow2 to raw).

If we are using COW, the libvirt pool.createXML method could be used with a backingStore element, which will appropriately create the new QCOW2 file with the backing file as the file in the image cache.

This has the additional benefit of paving the way for the simplification of the image cache manager – instead of having to run an external executable to check if an image is in the qcow2 format and has a backing store, we can simply check the backingStore element’s path subelement for each libvirt volume (this also makes the code less brittle, should we decide to support other formats with backing stores) [2].

A similar approach could be used with extract_snapshot – use createXMLFrom to duplicate the libvirt volume (the new XML we pass in can handle compression, etc) into a temporary directory pool.

In order to associate images with instances, the volumes in nova-disks-pool would have a name of the form {instance-uuid}_{name} (with name being “disk”, “kernel”, etc, depending on the name passed to the image creation method). This way, it still remains easy to find the disk image associated with a particular instance. This is the same name format used for the legacy LVM and RBD backends.

A configuration variable named [libvirt]use_storage_pools would enable or disable the storage pool functionality, and would be set to true by default. However, the legacy backends would be left in place to maintain the live upgrade functionality (e.g. Juno->Kilo). See the Other deployer impact section below for more information.

For the disk XML element in the domain element supplied to libvirt on instance creation, a type of volume can be supplied, with the <source> element specifying the pool name and volume name [3].

Alternatives

The setup described in this document calls for using a single storage pool for all VMs on a system.

When using a file-based backend, this would require storing disk images in a single directory (such as /var/lib/nova/instance/disks) instead of the current setup, where the disk images are stored in the instance directory (/var/lib/nova/instances/{instance-id}). This is due to the way that the libvirt dir storage pool works.

While it would be possible to create a new storage pool for each instance, this would only be applicable for file-based backends. Having different functionality between file-based backends and other backends would complicate the code and reduce the abstraction introduced by this blueprint.

Data model impact

None.

REST API impact

None.

Security impact

None.

Notifications impact

None.

Other end user impact

None.

Performance Impact

Since the createXMLFrom is actually intelligent about creating and copying image files (for instance, it calls qemu-img under the hood when appropriate), there should be no performance impact. As per what is mentioned in the Proposed change section, we would maintain current image cache functionality, including support for COW (via QCOW2), while paving the road for other file formats that libvirt supports as well.

Other deployer impact

For live migration/upgrade from OpenStack Juno to OpenStack Kilo, the legacy image backends (and support for them in Nova’s image cache) will be left in place for the next release (Kilo), but will be marked as deprecated. In the L release, the legacy backends will be removed (as well as support for them in the image cache manager).

when the deployer enables the [libvirt]use_storage_pools configuration options, there would be several effects:

First, Nova would check to see if the nova-image-cache-pool and nova-disks-pool already existed. If not, the nova-image-cache-pool storage pool would be created as a directory pool in the current image cache directory. Then, Nova would examine the current images type and attempt to use existing information to create the nova-disks-pool storage pool. The automated creation of the main storage pool would be a temporary measure to assist in the transitioning process; eventually (after L), this would be removed, since the configuration options for the legacy backends would also be removed. This lifts some of the burden from Nova on interacting with various storage backends – Nova would no longer have to have a multitude of configuration options for every storage backend it supported.

Secondly, all new instances would be created using the storage pool image backend. Any currently running instances would continue to use the legacy image backend.

During operations which allow the changing of libvirt XML, such as cold migrations, resizes, reboots, and live migrations [4], instances would be automatically transitioned to using the new system. This would allow deployers and users to move to the new system at their leisure, since they could either choose to bulk-restart the VMs themselves, or simply ask the users to do so when convinient. For instances still on the legacy system, a warning would be issued on compute node startup.

For “cold” operations (resizes, reboots, and cold migrations), disk images would be moved into the storage pool before the virtual machine was (re)started. For non-directory-based backends (LVM and RBD), no movement is necessary, since the name format is the same, and they already use a centralized location by their very nature.

Then, when Nova went to generate the new XML to boot the VM, the XML would point to the libvirt storage volume (in the case of a soft reboot, we would simply update the existing XML).

For live block migrations, we simply create a new, empty image in the storage pool, and let libvirt fill it up as part of the block migration. For shared storage live migrations, we can only transition if the image backend is Ceph, since there’s no reliable way to move a disk file into the storage pool while the VM is still running without losing data.

Developer impact

Currently, file-based images for a particular instance are stored in the instance directory (/var/lib/nova/instances/{instance-id}). In order to have one storage pool per compute node, libvirt’s directory-based storage pool would require all of the disk images to be stored in one directory, so the images themselves would no longer be in /var/lib/nova/instances/{instance-id}, but instead in something to the effect of /var/lib/nova/instance/disks.

Should it be desired to have different disk types (e.g. main disk vs swap) stored differently [5], we could simply create a pool for each type, and place the images into the appropriate pool based on their name. An advantage to using pools is that Nova doesn’t actually need to know the underlying details about the pool, only its name. Thus, if a deployer wanted to move a particular pool to a different location, device, etc, no XML changes would be needed, assuming the same pool name was kept.

Code that targets a specific backend type (such as LVM encryption, for instance) is still possible, since we can ask libvirt for the storage pool type.

Implementation

Assignee(s)

Primary assignee:

paul-carlton2

Other contributors:

None

Work Items

  1. Modify the code which downloads images from Glance into a cache to create a storage pool in the cache directory and refresh the cache when a new image is downloaded.

  2. Implement the new image backend and sections in the XML config builder to accept the volume type for disk elements, and make the image cache manager aware of how to check libvirt storage volumes for backing stores.

  3. Implement the functionality required to support transitional installations (detecting legacy backend use, adding code to migration and reboots to transition into new backend use).

  4. Subclass the new image backend for RBD support to allow it to be used with the new image backend.

Dependencies

No new libraries are required for this change. However, the XML changes discussed above require a libvirt version > 1.0.5 (the actual storage pools do not, however). While this is not strictly needed (as we can simply use the existing code for determining the correct XML for a given image), it does simplify the section of the code responsible for XML generation. Since we will most likely be increasing the minimum libvirt version for Mikata, however, this should not be problematic.

Testing

We will want to duplicate the existing tests for the various image backends to ensure that the new backend covers all of the existing functionality. Additionally, new tests should be introduced for:

  • the XML changes

  • storage pool management

  • migrating existing instances to the new backend and the supporting transitional functionality

Documentation Impact

We should warn about the deprecation of the legacy image backends, and note the change to the new backend. It should also be noted that migrations and cold resizes are the preferred method to transition existing instances to the new backend.

References