Nimble Storage volume driver¶
Nimble Storage fully integrates with the OpenStack platform through the Nimble Cinder driver, allowing a host to configure and manage Nimble Storage array features through Block Storage interfaces.
Support for iSCSI storage protocol is available with NimbleISCSIDriver Volume Driver class and Fibre Channel with NimbleFCDriver.
Support for the Liberty release and above is available from Nimble OS 2.3.8 or later.
Support for the Ocata release and above is available from Nimble OS 3.6 or later.
Nimble Storage Cinder driver does not support port binding with multiple interfaces on the same subnet due to existing limitation in os-brick. This is partially referenced in the bug https://bugs.launchpad.net/os-brick/+bug/1722432 but does not resolve for multiple software iscsi ifaces.
Supported operations¶
Create, delete, clone, attach, and detach volumes
Create and delete volume snapshots
Create a volume from a snapshot
Copy an image to a volume
Copy a volume to an image
Extend a volume
Get volume statistics
Manage and unmanage a volume
Enable encryption and default performance policy for a volume-type extra-specs
Force backup of an in-use volume.
Nimble Storage driver configuration¶
Update the file /etc/cinder/cinder.conf
with the given configuration.
In case of a basic (single back-end) configuration, add the parameters
within the [default]
section as follows.
[default]
san_ip = NIMBLE_MGMT_IP
san_login = NIMBLE_USER
san_password = NIMBLE_PASSWORD
use_multipath_for_image_xfer = True
volume_driver = NIMBLE_VOLUME_DRIVER
In case of multiple back-end configuration, for example, configuration which supports multiple Nimble Storage arrays or a single Nimble Storage array with arrays from other vendors, use the following parameters.
[default]
enabled_backends = Nimble-Cinder
[Nimble-Cinder]
san_ip = NIMBLE_MGMT_IP
san_login = NIMBLE_USER
san_password = NIMBLE_PASSWORD
use_multipath_for_image_xfer = True
volume_driver = NIMBLE_VOLUME_DRIVER
volume_backend_name = NIMBLE_BACKEND_NAME
In case of multiple back-end configuration, Nimble Storage volume type is created and associated with a back-end name as follows.
Note
Single back-end configuration users do not need to create the volume type.
$ openstack volume type create NIMBLE_VOLUME_TYPE
$ openstack volume type set --property volume_backend_name=NIMBLE_BACKEND_NAME NIMBLE_VOLUME_TYPE
This section explains the variables used above:
- NIMBLE_MGMT_IP
Management IP address of Nimble Storage array/group.
- NIMBLE_USER
Nimble Storage account login with minimum
power user
(admin) privilege if RBAC is used.- NIMBLE_PASSWORD
Password of the admin account for nimble array.
- NIMBLE_VOLUME_DRIVER
Use either cinder.volume.drivers.nimble.NimbleISCSIDriver for iSCSI or cinder.volume.drivers.nimble.NimbleFCDriver for Fibre Channel.
- NIMBLE_BACKEND_NAME
A volume back-end name which is specified in the
cinder.conf
file. This is also used while assigning a back-end name to the Nimble volume type.- NIMBLE_VOLUME_TYPE
The Nimble volume-type which is created from the CLI and associated with
NIMBLE_BACKEND_NAME
.Note
Restart the
cinder-api
,cinder-scheduler
, andcinder-volume
services after updating thecinder.conf
file.
Nimble driver extra spec options¶
The Nimble volume driver also supports the following extra spec options:
- ‘nimble:encryption’=’yes’
Used to enable encryption for a volume-type.
- ‘nimble:perfpol-name’=PERF_POL_NAME
PERF_POL_NAME is the name of a performance policy which exists on the Nimble array and should be enabled for every volume in a volume type.
- ‘nimble:multi-initiator’=’true’
Used to enable multi-initiator access for a volume-type.
- nimble:dedupe’=’true’
Used to enable dedupe support for a volume-type.
- ‘nimble:iops-limit’=IOPS_LIMIT
Used to set the IOPS_LIMIT between 256 and 4294967294 for all volumes created for this volume-type.
- ‘nimble:folder’=FOLDER_NAME
FOLDER_NAME is the name of the folder which exists on the Nimble array and should be enabled for every volume in a volume type
These extra-specs can be enabled by using the following command:
$ openstack volume type set --property KEY=VALUE VOLUME_TYPE
VOLUME_TYPE
is the Nimble volume type and KEY
and VALUE
are
the options mentioned above.
Configuration options¶
The Nimble storage driver supports these configuration options:
Configuration option = Default value |
Description |
---|---|
|
(String) Nimble Controller pool name |
|
(String) Nimble Subnet Label |
|
(String) Path to Nimble Array SSL certificate |
|
(Boolean) Whether to verify Nimble SSL Certificate |
Multipathing¶
In OpenStack environments where Cinder block device multipathing is desired there are a few things to consider.
Configuring mulitpathing varies by system depending on the environment. In a
scenario where solely Nimble devices are being created by Cinder, the
following /etc/multipath.conf
file may be used:
defaults {
user_friendly_names yes
find_multipaths no
}
blacklist {
devnode "^(ram|raw|loop|fd|md|dm-|sr|scd|st)[0-9]*"
devnode "^hd[a-z]"
device {
vendor ".*"
product ".*"
}
}
blacklist_exceptions {
device {
vendor "Nimble"
product "Server"
}
}
devices {
device {
vendor "Nimble"
product "Server"
path_grouping_policy group_by_prio
prio "alua"
hardware_handler "1 alua"
path_selector "service-time 0"
path_checker tur
features "1 queue_if_no_path"
no_path_retry 30
failback immediate
fast_io_fail_tmo 5
dev_loss_tmo infinity
rr_min_io_rq 1
rr_weight uniform
}
}
After making changes to /etc/multipath.conf
, the multipath subsystem needs
to be reconfigured:
# multipathd reconfigure
Tip
The latest best practices for Nimble devices can be found in the HPE Nimble Storage Linux Integration Guide found on https://infosight.hpe.com
Important
OpenStack Cinder is currently not compatible with the HPE Nimble Storage Linux Toolkit (NLT)
Nova needs to be configured to pickup the actual multipath device created on the host.
In /etc/nova/nova.conf
, add the following to the [libvirt]
section:
[libvirt]
volume_use_multipath = True
Note
In versions prior to Newton, the option was called iscsi_use_multipath
After editing the Nova configuration file, the nova-conductor
service
needs to be restarted.
Tip
Depending on which particular OpenStack distribution is being used, Nova may use a different configuration file than the default.
To validate that instances get properly connected to the multipath device, inspect the instance devices:
# virsh dumpxml <Instance ID | Instance Name | Instance UUID>