starlingx
/
docs
Code Issues Proposed changes

Merge "[Installation] All-In-One Duplex + Up to 4 Computes"

This commit is contained in:
Zuul 2019-02-05 20:31:59 +00:00 committed by Gerrit Code Review
parent a0ea45ed1b 3b3dcb197c
commit 984eac06a6
1 changed files with 287 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

287
doc/source/installation_guide/duplex.rst
View File

@ -1200,6 +1200,293 @@ excessive data-sync time). Use 'fm alarm-list' to confirm status.
| 2 | controller-1 | controller | unlocked | enabled | available |
+----+--------------+-------------+----------------+-------------+--------------+
-----------------------------------
Extending the Compute Node Capacity
-----------------------------------
You can add up to four Compute Nodes to the All-in-One Duplex Deployment.
**************************
Compute Hosts Installation
**************************
After initializing and configuring the two controllers, you can add up
to four additional compute hosts. To add a host, do the following:
^^^^^^^^^^^^^^^^^
Initializing Host
^^^^^^^^^^^^^^^^^
Power on the host. The following appears in the host console:
::
Waiting for this node to be configured.
Please configure the personality for this node from the
controller node in order to proceed.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Updating the Host Name and Personality
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
On Controller-0, acquire Keystone administrative privileges:
::
controller-0:~$ source /etc/nova/openrc
Wait for the Controller-0 to both discover the new host and to list that host
as UNKNOWN in the table:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-list
+----+--------------+-------------+----------------+-------------+--------------+
| id | hostname | personality | administrative | operational | availability |
+----+--------------+-------------+----------------+-------------+--------------+
| 1 | controller-0 | controller | unlocked | enabled | online |
| 2 | controller-1 | controller | unlocked | enabled | online |
| 3 | None | None | locked | disabled | offline |
+----+--------------+-------------+----------------+-------------+--------------+
Use the system host-update command to update the host personality attribute:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-update 3 personality=compute hostname=compute-0
See also: 'system help host-update'
Unless it is known that the host's configuration can support the
installation of more than one node, it is recommended that the
installation and configuration of each node be serialized. For example,
if the virtual disks for the entire cluster are hosted on the host's
root disk and that disk happens to be a single rotational type hard disk,
then the host cannot reliably support parallel node installation.
^^^^^^^^^^^^^^^
Monitoring Host
^^^^^^^^^^^^^^^
On Controller-0, you can monitor the installation progress by periodically
running the system host-show command for the host. Progress is
shown in the install_state field.
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-show <host> | grep install
| install_output | text |
| install_state | booting |
| install_state_info | None |
Wait while the host is installed, configured, and rebooted. Depending on
hardware, it could take up to 20 minutes for this process to complete.
When the reboot is complete, the host is reported as Locked, Disabled,
and Online.
^^^^^^^^^^^^^
Listing Hosts
^^^^^^^^^^^^^
You can use the system host-list command to list the hosts once the node
has been installed, configured, and rebooted:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-list
+----+--------------+-------------+----------------+-------------+--------------+
| id | hostname | personality | administrative | operational | availability |
+----+--------------+-------------+----------------+-------------+--------------+
| 1 | controller-0 | controller | unlocked | enabled | online |
| 2 | controller-1 | controller | unlocked | enabled | online |
| 3 | compute-0 | compute | locked | disabled | online |
+----+--------------+-------------+----------------+-------------+--------------+
**********************
Compute Host Provision
**********************
You must configure the network interfaces and the storage disks on a
host before you can unlock it. For each Compute Host, do the following:
On Controller-0, acquire Keystone administrative privileges:
::
controller-0:~$ source /etc/nova/openrc
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Provisioning Network Interfaces on a Compute Host
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
In order to identify hardware port names, types, and discovered
pci-addresses on Controller-0, list the host ports:
- **Only in Virtual Environment**: Ensure that the interface used is
one of those attached to the host bridge with model type "virtio" (i.e.
eth1000 and eth1001). The model type "e1000" emulated devices will
not work for provider networks:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-port-list compute-0
+--------------------------------------+---------+----------+--------------+...
| uuid | name | type | pci address |...
+--------------------------------------+---------+----------+--------------+...
| de9ec830-cf33-4a06-8985-cd3262f6ecd3 | enp2s1 | ethernet | 0000:02:01.0 |...
| 9def88fb-d871-4542-83e2-d4b6e35169a1 | enp2s2 | ethernet | 0000:02:02.0 |...
| b2e38da9-840f-446c-b466-ceb6479266f6 | eth1000 | ethernet | 0000:02:03.0 |...
| c1694675-643d-4ba7-b821-cd147450112e | eth1001 | ethernet | 0000:02:04.0 |...
+--------------------------------------+---------+----------+--------------+...
Use the following command to provision the data interface for Compute:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-if-modify -p providernet-a -c data compute-0 eth1000
^^^^^^^^^^^^^^^^^^^^^^^^^^^
VSwitch Virtual Environment
^^^^^^^^^^^^^^^^^^^^^^^^^^^
**Only in Virtual Environment**. If the compute node has more than four CPUs,
the system auto-configures the vswitch to use two cores. However, some virtual
environments do not properly support multi-queue, which is required in a
multi-CPU environment. Therefore, run the following command to reduce the
vswitch cores to one:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-cpu-modify compute-0 -f vswitch -p0 1
+--------------------------------------+-------+-----------+-------+--------+...
| uuid | log_c | processor | phy_c | thread |...
+--------------------------------------+-------+-----------+-------+--------+...
| 9d53c015-8dd5-4cd8-9abb-6f231fa773c2 | 0 | 0 | 0 | 0 |...
| 8620eaa7-19af-4ef7-a5aa-690ed57f01c7 | 1 | 0 | 1 | 0 |...
| f26598a5-8029-4d20-999e-2ec5d532d62e | 2 | 0 | 2 | 0 |...
| f7b3ce4a-10da-485d-873c-4e6988f7e9cf | 3 | 0 | 3 | 0 |...
+--------------------------------------+-------+-----------+-------+--------+...
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Provisioning Storage on a Compute Host
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Review the available disk space and capacity and then obtain the uuid(s) of
the physical disk(s) to be used for nova local:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-disk-list compute-0
+--------------------------------------+-----------+---------+---------+-------+------------+...
| uuid | device_no | device_ | device_ | size_ | available |...
| | de | num | type | gib | gib |...
+--------------------------------------+-----------+---------+---------+-------+------------+...
| 0ae45272-c9f4-4824-8405-f6c8946fda1e | /dev/sda | 2048 | HDD | 200.0 | 120.976 |...
| d751abfe-de57-4b23-b166-1d3d5b4d5ca6 | /dev/sdb | 2064 | HDD | 200.0 | 199.997 |...
+--------------------------------------+-----------+---------+---------+-------+------------+...
Use the following command to create the 'nova-local' local volume group:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-lvg-add compute-0 nova-local
+-----------------------+-------------------------------------------------------------------+
| Property | Value |
+-----------------------+-------------------------------------------------------------------+
| lvm_vg_name | nova-local |
| vg_state | adding |
| uuid | 78e84a57-c40e-438c-97a4-49ebec6394d1 |
| ihost_uuid | 6f2de9b6-c55c-47b0-b40e-aff47f1e1bda |
| lvm_vg_access | None |
| lvm_max_lv | 0 |
| lvm_cur_lv | 0 |
| lvm_max_pv | 0 |
| lvm_cur_pv | 0 |
| lvm_vg_size_gib | 0.0 |
| lvm_vg_avail_size_gib | 0.0 |
| lvm_vg_total_pe | 0 |
| lvm_vg_free_pe | 0 |
| created_at | 2019-01-15T12:31:45.796538+00:00 |
| updated_at | None |
| parameters | {u'concurrent_disk_operations': 2, u'instance_backing': u'image'} |
+-----------------------+-------------------------------------------------------------------+
Use the following command to create a disk partition to add to the volume
group based on the uuid of the physical disk:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-pv-add compute-0 nova-local d751abfe-de57-4b23-b166-1d3d5b4d5ca6
+--------------------------+--------------------------------------------+
| Property | Value |
+--------------------------+--------------------------------------------+
| uuid | bba20f20-81fc-40a2-bf1e-a2fae849625b |
| pv_state | adding |
| pv_type | disk |
| disk_or_part_uuid | d751abfe-de57-4b23-b166-1d3d5b4d5ca6 |
| disk_or_part_device_node | /dev/sdb |
| disk_or_part_device_path | /dev/disk/by-path/pci-0000:00:1f.2-ata-2.0 |
| lvm_pv_name | /dev/sdb |
| lvm_vg_name | nova-local |
| lvm_pv_uuid | None |
| lvm_pv_size_gib | 0.0 |
| lvm_pe_total | 0 |
| lvm_pe_alloced | 0 |
| ihost_uuid | 6f2de9b6-c55c-47b0-b40e-aff47f1e1bda |
| created_at | 2019-01-15T12:32:40.535160+00:00 |
| updated_at | None |
+--------------------------+--------------------------------------------+
^^^^^^^^^^^^^^^^^^^^^^^^
Unlocking a Compute Host
^^^^^^^^^^^^^^^^^^^^^^^^
On Controller-0, use the system host-unlock command to unlock the
Compute node:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-unlock compute-0
Wait while the Compute node is rebooted and re-configured. Depending on
hardware, it can take up to 10 minutes for the reboot to complete. Once
the reboot is complete, the nodes Availability State reports as "In-Test"
and is followed by unlocked/enabled.
-------------------
System Health Check
-------------------
***********************
Listing StarlingX Nodes
***********************
On Controller-0, after a few minutes, all nodes are reported as
Unlocked, Enabled, and Available:
::
[wrsroot@controller-0 ~(keystone_admin)]$ system host-list
+----+--------------+-------------+----------------+-------------+--------------+
| id | hostname | personality | administrative | operational | availability |
+----+--------------+-------------+----------------+-------------+--------------+
| 1 | controller-0 | controller | unlocked | enabled | available |
| 2 | controller-1 | controller | unlocked | enabled | available |
| 3 | compute-0 | compute | unlocked | enabled | available |
+----+--------------+-------------+----------------+-------------+--------------+
*****************
System Alarm List