cinder/doc/source/configuration/block-storage/drivers/dell-emc-unity-driver.rst
Ryan Liang b44721dfac [Unity] Add force detach support
Add support to force detach a volume from all hosts on Unity.

Closes-bug: #1741831
Change-Id: I5f0f3d474b00f3c1f742c1bfa6877911415c32f9
2018-01-17 17:02:35 +08:00

374 lines
10 KiB
ReStructuredText

=====================
Dell EMC Unity driver
=====================
Unity driver has been integrated in the OpenStack Block Storage project since
the Ocata release. The driver is built on the top of Block Storage framework
and a Dell EMC distributed Python package
`storops <https://pypi.python.org/pypi/storops>`_.
Prerequisites
~~~~~~~~~~~~~
+-------------------+----------------+
| Software | Version |
+===================+================+
| Unity OE | 4.1.X |
+-------------------+----------------+
| OpenStack | Ocata |
+-------------------+----------------+
| storops | 0.4.2 or newer |
+-------------------+----------------+
Supported operations
~~~~~~~~~~~~~~~~~~~~
- Create, delete, attach, and detach volumes.
- Create, list, and delete volume snapshots.
- Create a volume from a snapshot.
- Copy an image to a volume.
- Create an image from a volume.
- Clone a volume.
- Extend a volume.
- Migrate a volume.
- Get volume statistics.
- Efficient non-disruptive volume backup.
Driver configuration
~~~~~~~~~~~~~~~~~~~~
.. note:: The following instructions should all be performed on Black Storage
nodes.
#. Install `storops` from pypi:
.. code-block:: console
# pip install storops
#. Add the following content into ``/etc/cinder/cinder.conf``:
.. code-block:: ini
[DEFAULT]
enabled_backends = unity
[unity]
# Storage protocol
storage_protocol = iSCSI
# Unisphere IP
san_ip = <SAN IP>
# Unisphere username and password
san_login = <SAN LOGIN>
san_password = <SAN PASSWORD>
# Volume driver name
volume_driver = cinder.volume.drivers.dell_emc.unity.Driver
# backend's name
volume_backend_name = Storage_ISCSI_01
.. note:: These are minimal options for Unity driver, for more options,
see `Driver options`_.
.. note:: (**Optional**) If you require multipath based data access, perform
below steps on both Block Storage and Compute nodes.
#. Install ``sysfsutils``, ``sg3-utils`` and ``multipath-tools``:
.. code-block:: console
# apt-get install multipath-tools sg3-utils sysfsutils
#. (Required for FC driver in case `Auto-zoning Support`_ is disabled) Zone the
FC ports of Compute nodes with Unity FC target ports.
#. Enable Unity storage optimized multipath configuration:
Add the following content into ``/etc/multipath.conf``
.. code-block:: vim
blacklist {
# Skip the files uner /dev that are definitely not FC/iSCSI devices
# Different system may need different customization
devnode "^(ram|raw|loop|fd|md|dm-|sr|scd|st)[0-9]*"
devnode "^hd[a-z][0-9]*"
devnode "^cciss!c[0-9]d[0-9]*[p[0-9]*]"
# Skip LUNZ device from VNX/Unity
device {
vendor "DGC"
product "LUNZ"
}
}
defaults {
user_friendly_names no
flush_on_last_del yes
}
devices {
# Device attributed for EMC CLARiiON and VNX/Unity series ALUA
device {
vendor "DGC"
product ".*"
product_blacklist "LUNZ"
path_grouping_policy group_by_prio
path_selector "round-robin 0"
path_checker emc_clariion
features "0"
no_path_retry 12
hardware_handler "1 alua"
prio alua
failback immediate
}
}
#. Restart the multipath service:
.. code-block:: console
# service multipath-tools restart
#. Enable multipath for image transfer in ``/etc/cinder/cinder.conf``.
.. code-block:: ini
use_multipath_for_image_xfer = True
Restart the ``cinder-volume`` service to load the change.
#. Enable multipath for volume attache/detach in ``/etc/nova/nova.conf``.
.. code-block:: ini
[libvirt]
...
volume_use_multipath = True
...
#. Restart the ``nova-compute`` service.
Driver options
~~~~~~~~~~~~~~
.. include:: ../../tables/cinder-dell_emc_unity.inc
FC or iSCSI ports option
------------------------
Specify the list of FC or iSCSI ports to be used to perform the IO. Wild card
character is supported.
For iSCSI ports, use the following format:
.. code-block:: ini
unity_io_ports = spa_eth2, spb_eth2, *_eth3
For FC ports, use the following format:
.. code-block:: ini
unity_io_ports = spa_iom_0_fc0, spb_iom_0_fc0, *_iom_0_fc1
List the port ID with the :command:`uemcli` command:
.. code-block:: console
$ uemcli /net/port/eth show -output csv
...
"spa_eth2","SP A Ethernet Port 2","spa","file, net, iscsi", ...
"spb_eth2","SP B Ethernet Port 2","spb","file, net, iscsi", ...
...
$ uemcli /net/port/fc show -output csv
...
"spa_iom_0_fc0","SP A I/O Module 0 FC Port 0","spa", ...
"spb_iom_0_fc0","SP B I/O Module 0 FC Port 0","spb", ...
...
Live migration integration
~~~~~~~~~~~~~~~~~~~~~~~~~~
It is suggested to have multipath configured on Compute nodes for robust data
access in VM instances live migration scenario. Once ``user_friendly_names no``
is set in defaults section of ``/etc/multipath.conf``, Compute nodes will use
the WWID as the alias for the multipath devices.
To enable multipath in live migration:
.. note:: Make sure `Driver configuration`_ steps are performed before
following steps.
#. Set multipath in ``/etc/nova/nova.conf``:
.. code-block:: ini
[libvirt]
...
volume_use_multipath = True
...
Restart `nova-compute` service.
#. Set ``user_friendly_names no`` in ``/etc/multipath.conf``
.. code-block:: text
...
defaults {
user_friendly_names no
}
...
#. Restart the ``multipath-tools`` service.
Thin and thick provisioning
~~~~~~~~~~~~~~~~~~~~~~~~~~~
Only thin volume provisioning is supported in Unity volume driver.
QoS support
~~~~~~~~~~~
Unity driver supports ``maxBWS`` and ``maxIOPS`` specs for the back-end
consumer type. ``maxBWS`` represents the ``Maximum IO/S`` absolute limit,
``maxIOPS`` represents the ``Maximum Bandwidth (KBPS)`` absolute limit on the
Unity respectively.
Auto-zoning support
~~~~~~~~~~~~~~~~~~~
Unity volume driver supports auto-zoning, and share the same configuration
guide for other vendors. Refer to :ref:`fc_zone_manager`
for detailed configuration steps.
Solution for LUNZ device
~~~~~~~~~~~~~~~~~~~~~~~~
The EMC host team also found LUNZ on all of the hosts, EMC best practice is to
present a LUN with HLU 0 to clear any LUNZ devices as they can cause issues on
the host. See KB `LUNZ Device <https://support.emc.com/kb/463402>`_.
To workaround this issue, Unity driver creates a `Dummy LUN` (if not present),
and adds it to each host to occupy the `HLU 0` during volume attachment.
.. note:: This `Dummy LUN` is shared among all hosts connected to the Unity.
Efficient non-disruptive volume backup
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The default implementation in Block Storage for non-disruptive volume backup is
not efficient since a cloned volume will be created during backup.
An effective approach to backups is to create a snapshot for the volume and
connect this snapshot to the Block Storage host for volume backup.
IPv6 support
~~~~~~~~~~~~
This driver can support IPv6-based control path and data path.
For control path, please follow below steps:
- Enable Unity's Unipshere IPv6 address.
- Configure the IPv6 network to make sure that cinder node can access Unishpere
via IPv6 address.
- Change Cinder config file ``/etc/cinder/cinder.conf``. Make the ``san_ip``
as Unisphere IPv6 address. For example, ``san_ip = [fd99:f17b:37d0::100]``.
- Restart the Cinder service to make new configuration take effect.
**Note**: The IPv6 support on control path depends on the fix of cpython
`bug 32185 <https://bugs.python.org/issue32185>`__. Please make sure your
Python's version includes this bug's fix.
For data path, please follow below steps:
- On Unity, Create iSCSI interface with IPv6 address.
- Configure the IPv6 network to make sure that you can ``ping``
the Unity's iSCSI IPv6 address from the Cinder node.
- If you create a volume using Cinder and attach it to a VM,
the connection between this VM and volume will be IPv6-based iSCSI.
Force detach volume from all hosts
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The user could use `os-force_detach` action to detach a volume from all its
attached hosts.
For more detail, please refer to
https://developer.openstack.org/api-ref/block-storage/v2/?expanded=force-detach-volume-detail#force-detach-volume
Troubleshooting
~~~~~~~~~~~~~~~
To troubleshoot a failure in OpenStack deployment, the best way is to
enable verbose and debug log, at the same time, leverage the build-in
`Return request ID to caller
<https://specs.openstack.org/openstack/openstack-specs/specs/return-request-id.html>`_
to track specific Block Storage command logs.
#. Enable verbose log, set following in ``/etc/cinder/cinder.conf`` and restart
all Block Storage services:
.. code-block:: ini
[DEFAULT]
...
debug = True
verbose = True
...
If other projects (usually Compute) are also involved, set `debug`
and ``verbose`` to ``True``.
#. use ``--debug`` to trigger any problematic Block Storage operation:
.. code-block:: console
# cinder --debug create --name unity_vol1 100
You will see the request ID from the console, for example:
.. code-block:: console
DEBUG:keystoneauth:REQ: curl -g -i -X POST
http://192.168.1.9:8776/v2/e50d22bdb5a34078a8bfe7be89324078/volumes -H
"User-Agent: python-cinderclient" -H "Content-Type: application/json" -H
"Accept: application/json" -H "X-Auth-Token:
{SHA1}bf4a85ad64302b67a39ad7c6f695a9630f39ab0e" -d '{"volume": {"status":
"creating", "user_id": null, "name": "unity_vol1", "imageRef": null,
"availability_zone": null, "description": null, "multiattach": false,
"attach_status": "detached", "volume_type": null, "metadata": {},
"consistencygroup_id": null, "source_volid": null, "snapshot_id": null,
"project_id": null, "source_replica": null, "size": 10}}'
DEBUG:keystoneauth:RESP: [202] X-Compute-Request-Id:
req-3a459e0e-871a-49f9-9796-b63cc48b5015 Content-Type: application/json
Content-Length: 804 X-Openstack-Request-Id:
req-3a459e0e-871a-49f9-9796-b63cc48b5015 Date: Mon, 12 Dec 2016 09:31:44 GMT
Connection: keep-alive
#. Use commands like ``grep``, ``awk`` to find the error related to the Block
Storage operations.
.. code-block:: console
# grep "req-3a459e0e-871a-49f9-9796-b63cc48b5015" cinder-volume.log