Pure Storage - pure¶
Pure Storage is a software-defined storage solution. It offers the consumption of redundant block storage across the network.
LXD supports connecting to Pure Storage storage clusters through three protocols: iSCSI, NVMe/TCP, or FC. In addition, Pure Storage offers copy-on-write snapshots, thin provisioning, and other features.
To use Pure Storage with LXD requires a Pure Storage API version of at least 2.21, corresponding to a minimum Purity//FA version of 6.4.2.
Additionally, ensure that the required kernel modules for the selected protocol are installed on your host system.
For iSCSI, the iSCSI CLI named iscsiadm needs to be installed in addition to the required kernel modules.
For Fibre Channel, the scsi_transport_fc kernel module is required, along with a Fibre Channel HBA that is zoned to the array.
Fibre Channel volumes are always accessed through multipath, therefore multipath-tools must be installed and the multipath daemon must be running.
Terminology¶
Each storage pool created in LXD using a Pure Storage driver represents a Pure Storage pod, which is an abstraction that groups multiple volumes under a specific name. One benefit of using Pure Storage pods is that they can be linked with multiple Pure Storage arrays to provide additional redundancy.
LXD creates volumes within a pod that is identified by the storage pool name.
When the first volume needs to be mapped to a specific LXD host, a corresponding Pure Storage host is created with the name of the LXD host and a suffix of the used protocol.
For example, if the LXD host is host01 and the mode is nvme/tcp, the resulting Pure Storage host would be host01-nvme-tcp.
Likewise, in scsi/fc mode the resulting Pure Storage host would be host01-scsi-fc, because a slash (/) is not valid in a host name and is therefore replaced with a hyphen (-).
The Pure Storage host is then connected with the required volumes, to allow attaching and accessing volumes from the LXD host. The created Pure Storage host is automatically removed once there are no volumes connected to it.
The pure driver in LXD¶
The pure driver in LXD uses Pure Storage volumes for custom storage volumes, instances, and snapshots.
All created volumes are thin-provisioned block volumes. If required (for example, for containers and custom file system volumes), LXD formats the volume with a desired file system.
LXD expects Pure Storage to be pre-configured with a specific service (e.g. iSCSI) on network interfaces whose address is provided during storage pool configuration.
This does not apply in scsi/fc mode, because Fibre Channel target ports are not network interfaces.
Furthermore, LXD assumes that it has full control over the Pure Storage pods it manages.
Therefore, you should never maintain any volumes in Pure Storage pods that are not owned by LXD because LXD might disconnect or even delete them.
This driver behaves differently than some of the other drivers in that it provides remote storage. As a result, and depending on the internal network, storage access might be a bit slower compared to local storage. On the other hand, using remote storage has significant advantages in a cluster setup: all cluster members have access to the same storage pools with the exact same contents, without the need to synchronize them.
When creating a new storage pool using the pure driver in either iscsi or nvme/tcp mode, LXD automatically discovers the array’s qualified name and target address (portal).
In scsi/fc mode, LXD instead discovers the online Fibre Channel target ports through the local host bus adapter, because Fibre Channel targets are identified by WWPN rather than by network address.
Consequently, pure.target has no effect in scsi/fc mode, and, instead, fabric zoning determines which targets are reachable.
Upon successful discovery, LXD attaches all volumes that are connected to the Pure Storage host that is associated with a specific LXD server. Pure Storage hosts and volume connections are fully managed by LXD.
Volume snapshots are also supported by Pure Storage. When a volume with at least one snapshot is copied, LXD sequentially creates snapshots on the destination volume from snapshots on the source volume. Each snapshot is associated with a parent volume and cannot be directly attached to the host; therefore, when a snapshot is exported, LXD creates a temporary volume behind the scenes. This volume is attached to the LXD host and removed once the operation is complete. Finally, once all snapshots are copied, the source volume is copied into the destination volume.
Volume names¶
Due to a limitation in Pure Storage, volume names cannot exceed 63 characters.
Therefore, the driver uses the volume’s volatile.uuid to generate a shorter volume name.
For example, a UUID 5a2504b0-6a6c-4849-8ee7-ddb0b674fd14 is first trimmed of any hyphens (-), resulting in the string 5a2504b06a6c48498ee7ddb0b674fd14.
To distinguish volume types and snapshots, special identifiers are prepended and appended to the volume names, as depicted in the table below:
Type |
Identifier |
Example |
|---|---|---|
Container |
|
|
Virtual machine |
|
|
Image (ISO) |
|
|
Custom volume |
|
|
Snapshot |
|
|
Limitations¶
The pure driver has the following limitations:
- Volume size constraints
Minimum volume size (quota) is
1MiBand must be a multiple of512B. If the requested size does not meet these conditions, LXD automatically rounds it up to the nearest valid value.- Snapshots cannot be mounted
Snapshots cannot be mounted directly to the host. Instead, a temporary volume must be created to access the snapshot’s contents. For internal operations, such as copying instances or exporting snapshots, LXD handles this automatically.
- Sharing the Pure Storage storage pool between multiple LXD installations
Sharing a Pure Storage array between multiple LXD installations is possible provided that installations use distinct storage pool names. Storage pools are implemented as Pods on the array and pod names have to be unique.
- Recovering Pure Storage storage pools
Recovery of Pure Storage storage pools using
lxd recoveris currently not supported.- Only one Fibre Channel initiator is used
In
scsi/fcmode, LXD registers a single initiator WWPN with the array: the first one the host reports. A host can have several Fibre Channel HBA cards, each card can provide several ports, and every port has its own WWPN. All initiators except the first are left unregistered, and the array does not present volumes to them. Redundancy is therefore limited to the paths reachable through the registered initiator, and its loss means that the pool can no longer be accessed even if the host has other Fibre Channel ports. Which initiator is registered depends on the order in which they are enumerated by the host, which is not guaranteed to be stable across reboots. Zone the fabric so that the registered initiator reaches every array target port that should be used.
Configuration options¶
The following configuration options are available for storage pools that use the pure driver, as well as storage volumes in these pools.
Storage pool configuration¶
| Key: | pure.api.token |
| Type: | string |
API authorization token for Pure Storage gateway. Must have array_admin role to give LXD full control over managed storage pools (Pure Storage pods).
| Key: | pure.gateway.verify |
| Type: | bool |
| Default: |
|
| Key: | pure.mode |
| Type: | string |
| Default: |
|
The mode to use to map Pure Storage volumes to the local server.
Supported values are iscsi, nvme/tcp, and scsi/fc.
| Key: | pure.target |
| Type: | string |
| Default: | all available targets |
A comma-separated list of target addresses. If empty, LXD discovers and connects to all available targets. Otherwise, it only connects to the specified addresses.
This option has no effect when pure.mode is set to scsi/fc, because Fibre Channel targets are identified by WWPN rather than by network address.
| Key: | rsync.bwlimit |
| Type: | string |
| Default: |
|
| Scope: | global |
When rsync must be used to transfer storage entities, this option specifies the upper limit
to be placed on the socket I/O.
| Key: | rsync.compression |
| Type: | bool |
| Default: |
|
| Scope: | global |
| Key: | volume.size |
| Type: | string |
| Default: |
|
Default Pure Storage volume size rounded to 512B. The minimum size is 1MiB.
Tip
In addition to these configurations, you can also set default values for the storage volume configurations. See Configure default values for storage volumes.
Storage volume configuration¶
| Key: | block.filesystem |
| Type: | string |
| Default: | same as |
| Condition: | block-based volume with content type |
Valid options: btrfs, ext4, xfs
If not set, ext4 is assumed.
| Key: | block.mount_options |
| Type: | string |
| Default: | same as |
| Condition: | block-based volume with content type |
| Key: | security.shifted |
| Type: | bool |
| Default: | same as |
| Condition: | custom volume |
| Scope: | global |
Enable this option to allow the volume to be attached to multiple isolated instances.
| Key: | security.unmapped |
| Type: | bool |
| Default: | same as |
| Condition: | custom volume |
| Scope: | global |
| Key: | size |
| Type: | string |
| Default: | same as |
Default Pure Storage volume size rounded to 512B. The minimum size is 1MiB.
| Key: | snapshots.expiry |
| Type: | string |
| Default: | same as |
| Condition: | custom volume |
| Scope: | global |
Specify an expression like 1M 2H 3d 4w 5m 6y.
| Key: | snapshots.pattern |
| Type: | string |
| Default: | same as |
| Condition: | custom volume |
| Scope: | global |
You can specify a naming template for scheduled snapshots and unnamed snapshots.
The snapshots.pattern option takes a Pongo2 template string to format the snapshot name.
To add a time stamp to the snapshot name, use the Pongo2 context variable creation_date.
Make sure to format the date in your template string to avoid forbidden characters in the snapshot name.
For example, set snapshots.pattern to {{ creation_date|date:'2006-01-02_15-04-05' }} to name the snapshots after their time of creation, down to the precision of a second.
Another way to avoid name collisions is to use the placeholder %d in the pattern.
If no matching snapshots exist, the placeholder is replaced with 0.
Otherwise, it is replaced with the next snapshot index, which is one higher than the highest existing matching snapshot index.
| Key: | snapshots.schedule |
| Type: | string |
| Default: | same as |
| Condition: | custom volume |
| Scope: | global |
Specify either a cron expression (<minute> <hour> <dom> <month> <dow>), a comma-separated list of schedule aliases (@hourly, @daily, @midnight, @weekly, @monthly, @annually, @yearly), or leave empty to disable automatic snapshots (the default).
| Key: | volatile.devlxd.owner |
| Type: | string |
| Default: | DevLXD owner identity ID |
| Scope: | global |
| Key: | volatile.idmap.last |
| Type: | string |
| Condition: | filesystem |
| Key: | volatile.idmap.next |
| Type: | string |
| Condition: | filesystem |