Bug 1802188 - overcloud-minimal ceph deployment on OSP16
Summary: overcloud-minimal ceph deployment on OSP16
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: Red Hat OpenStack
Classification: Red Hat
Component: documentation
Version: 16.0 (Train)
Hardware: Unspecified
OS: Unspecified
medium
low
Target Milestone: ---
: ---
Assignee: Laura Marsh
QA Contact: RHOS Documentation Team
URL:
Whiteboard:
Depends On:
Blocks: 1805432
TreeView+ depends on / blocked
 
Reported: 2020-02-12 14:47 UTC by Rafael Urena
Modified: 2020-03-05 15:02 UTC (History)
6 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
: 1805432 (view as bug list)
Environment:
Last Closed: 2020-03-05 15:02:17 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)

Description Rafael Urena 2020-02-12 14:47:15 UTC
Description of problem:
I am deploying osp16 with ceph and wanted to take advantage of the overcloud-minimal image and found that the image is lacking openvswitch package.

Version-Release number of selected component (if applicable):


How reproducible:
every time

Steps to Reproduce:
1. modify so that ceph nodes use minimal image following https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/16.0/html/director_installation_and_usage/creating-a-basic-overcloud-with-cli-tools#using-the-overcloud-minimal-image-to-avoid-using-a-Red-Hat-subscription-entitlement
2. deploy overcloud

Actual results:
deploy will fail with 
fatal: [cephstorage-0]: FAILED! => {"changed": false, "failures": ["No package openvswitch available."], "msg": "Failed to install some of the spec
ified packages", "rc": 1, "results": []}


Expected results:
Ceph nodes deploy and the overcloud deployment completes

Additional info:
I added the package to the image, updated overcloud-minimal image and tried a clean deploy. Now i am getting:
fatal: [osp-test1977-cephstorage-0]: FAILED! => {"changed": false, "msg": "Unable to start service systemd-modules-load.service: Job for systemd-modules-load.se
rvice failed because the control process exited with error code.\nSee \"systemctl status systemd-modules-load.service\" and \"journalctl -xe\" for details.\n"}
 
I would attach the sos from the ceph nodes but while i can ssh when they are created, when heat finishes, when this error occurs i try to ssh and i am instantly disconnected:

[stack@undercloud ~]$ ssh heat-admin.0.125
Unable to get valid context for heat-admin
This system is not registered to Red Hat Insights. See https://cloud.redhat.com/
To register this system, run: insights-client --register

Last login: Tue Feb 11 18:59:45 2020 from 10.10.0.10
Connection to 10.10.0.125 closed.
[stack@undercloud ~]$ 

Not sure why.

Comment 1 Jon Schlueter 2020-02-13 14:19:37 UTC
overcloud-minimal intentionally is lacking openvswitch by design, the deployment code that triggered this situation will need to be looked at if it is trying to install openvswitch.

Comment 2 John Fulton 2020-02-13 14:54:44 UTC
By using the overcloud minimal image you are avoiding the need to use OpenStack subscriptions. The OpenStack subscription contains OVS and thus OVS is not available in the minimal image by design. It is not necessary to use OVS configure CephStorage nodes as the networking options provided by RHEL alone are sufficient and director has the ability to configure them.

- use not ovs_bonds, but use linux_bonds:

https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html-single/advanced_overcloud_customization/index#linux-bonding-options

- use not ovs_bridge, but use linux_bridge:

 https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html-single/advanced_overcloud_customization/index#network-interface-reference

For further clarification, I'm not recommending that you use the linux_bridge Neutron plugin. Neutron plugins are not necessary for CephStorage nodes. If you were to use HyperConverged nodes, then that would be a different scenario and the minimal image is not sufficient for that node type.

Comment 3 John Fulton 2020-02-13 14:58:07 UTC
Rafael Urena: to resolve your issue please update your Heat Templates as recommended in my previous comment. This is a usage issue.

I am converting this to a docbug since our documentation on using the minimal image doesn't seem state the recommendation that is in my previous comment.

Comment 4 Dan Sneddon 2020-02-13 20:24:37 UTC
(In reply to John Fulton from comment #2)

> 
> - use not ovs_bridge, but use linux_bridge:
> 
>  https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/
> html-single/advanced_overcloud_customization/index#network-interface-
> reference
> 
> For further clarification, I'm not recommending that you use the
> linux_bridge Neutron plugin. Neutron plugins are not necessary for
> CephStorage nodes. If you were to use HyperConverged nodes, then that would
> be a different scenario and the minimal image is not sufficient for that
> node type.

Actually, bridges are not required and in general should not be used on Ceph nodes. Bridges are required on Compute nodes for attaching VMs to tenant or provider VLANs, and OVS bridges are required when using OVS bonding. However, a bridge adds another layer between the node and the network that shouldn't be required for Ceph nodes, and for maximum performance you should just configure interfaces, VLANs, and bonds on Ceph nodes with no bridges.

So simply using a network interface configuration template with no bridges is the ideal configuration for Ceph nodes, and using Linux kernel bonds means that OVS is not required.


Note You need to log in before you can comment on or make changes to this bug.