extending comment #0:

3.2. Upgrading the Director[1] says to subscribe osp9 repos. Where as step "3.4.3. Upgrading Keystone" will fail if openstack-puppet-modules package is updated to "openstack-puppet-modules-8.1.8-2.el7ost.noarch" which is shipped with osp9 repo.

https://bugzilla.redhat.com/show_bug.cgi?id=1365691#c4

Hence, enable osp9 repo only to install os-collect-config-0.1.37-6.el7ost . Once its installed disable it and subscribe to osp8.rpeo


_______________________________________

3.4.2. Installing Aodh:

Before running the step make sure that the overcloud controller nodes are patched with the latest version :  os-collect-config-0.1.37-6.el7ost, Or one will hit the following bug:

https://bugzilla.redhat.com/show_bug.cgi?id=1350489

Hence enable osp9 repos on each of the overcloud controller nodes an update it manually. If the overcloud nodes are registered to RHN, then install RHEA-2016:1599-1

$yum update -y --advisory=RHEA-2016:1599-1 advisory  
    Or
$yum update -y os-collect-config


[1] https://access.redhat.com/documentation/en/red-hat-openstack-platform/9/single/upgrading-red-hat-openstack-platform#sect-Major-Updating_Director_Packages

Hi,

(In reply to Pratik Pravin Bandarkar from comment #2)
> extending comment #0:
> 
> 3.2. Upgrading the Director[1] says to subscribe osp9 repos. Where as step
> "3.4.3. Upgrading Keystone" will fail if openstack-puppet-modules package is
> updated to "openstack-puppet-modules-8.1.8-2.el7ost.noarch" which is shipped
> with osp9 repo.
> 
> https://bugzilla.redhat.com/show_bug.cgi?id=1365691#c4
> 
> Hence, enable osp9 repo only to install os-collect-config-0.1.37-6.el7ost .
> Once its installed disable it and subscribe to osp8.rpeo

This bz has been closed as notabug, so I don't think this is relevant.

> 
> 
> _______________________________________
> 
> 3.4.2. Installing Aodh:
> 
> Before running the step make sure that the overcloud controller nodes are
> patched with the latest version :  os-collect-config-0.1.37-6.el7ost, Or one
> will hit the following bug:
> 
> https://bugzilla.redhat.com/show_bug.cgi?id=1350489
> 
> Hence enable osp9 repos on each of the overcloud controller nodes an update
> it manually. If the overcloud nodes are registered to RHN, then install
> RHEA-2016:1599-1
> 
> $yum update -y --advisory=RHEA-2016:1599-1 advisory  
>     Or
> $yum update -y os-collect-config

I think this should be added.  We never know in which version the user is starting his/her upgrade.

> 
> 
> [1]
> https://access.redhat.com/documentation/en/red-hat-openstack-platform/9/
> single/upgrading-red-hat-openstack-platform#sect-Major-
> Updating_Director_Packages

Hi,

(In reply to VIKRANT from comment #0)
> Description of problem:
> 
> Some suggestions for guide -
> https://access.redhat.com/documentation/en/red-hat-openstack-platform/9/
> single/upgrading-red-hat-openstack-platform/#sect-Important_Pre-Upgrade_Notes
> 
> 1) Section "2.2. Updating the Overcloud Images" it's better to mention the
> user that what changes will happen after the updating the existing images or
> we can add a note to take the output of glance image-list before updating
> the existing images so that user can compare the output of commands. 
> 
> 2) When I used below command it didn't work for me.
> 
> openstack overcloud image upload --update-existing --image-path ~/images/.
> 
> I used following command :
> 
> cd images/
> openstack overcloud image upload --update-existing --image-path .

Hum, bad.  This should work.  Going to check this one when I have a osp9 env ready.  Meanwhile, let's take the safe route and change the documentation.

> 
> 3) Section "3.2. Upgrading the Director" we should replace the heat
> stack-list command with "openstack stack list", as heat stack commands give
> WARNING Message now.

ack.

> 
> 4) Section "3.4.1. Pre-Upgrade Notes for the Overcloud" The default timezone
> for Red Hat OpenStack Platform 9 is now UTC. The default timezone for Red
> Hat OpenStack Platform 7 was EST."  It's also better to mention about the
> default TZ in RHEL OSP 8 as well.

ack, change "OpenStack Platform 7" to "OpenStack Platform 8".


> 
> b) Again in same section "glanceregistry endpoints are removed from
> enable-tls.yaml" after the upgrade. In RHEL OSP 8 setup deployed with TLS,
> glanceregistry endpoints where also present in enable-tls.yaml file but
> those are not present in RHEL OSP 9.
> 
> ~~~
> # diff -Nary
> /usr/share/openstack-tripleo-heat-templates/environments/enable-tls.yaml
> /home/stack/baremetal003/enable-tls.yaml | grep "Glance"
>     GlanceAdmin: {protocol: 'http', port: '9292', host: 'IP_A       
> GlanceAdmin: {protocol: 'http', port: '9292', host: 'IP_A
>     GlanceInternal: {protocol: 'http', port: '9292', host: 'I       
> GlanceInternal: {protocol: 'http', port: '9292', host: 'I
>     GlancePublic: {protocol: 'https', port: '13292', host: 'C       
> GlancePublic: {protocol: 'https', port: '13292', host: 'C
>     GnocchiAdmin: {protocol: 'http', port: '8041', host: 'IP_ |       
> GlanceRegistryAdmin: {protocol: 'http', port: '9191', hos
>     GnocchiInternal: {protocol: 'http', port: '8041', host: ' |       
> GlanceRegistryInternal: {protocol: 'http', port: '9191', 
>     GnocchiPublic: {protocol: 'https', port: '13041', host: ' |       
> GlanceRegistryPublic: {protocol: 'https', port: '9191', h
> ~~~

Hum not sure this should be user facing.  From the commit:

Remove GlanceRegistry from EndpointMap

We don't need an endpoint for the glance-registry service, that is
used by glance-api when needed and is not meant to be user-facing.

https://review.openstack.org/#/c/289812/

> 
> c) It's better to explicitly add the in pre-upgrade notes that if we need to
> configure ceph as backend for Gnocchi then we should add this in
> storage-environment.yaml file. 

You mean the glanceregistry parameters ?


> 
> d) During the upgrade I was confused from where should I get the value of
> other keys because I didn't use parameters during the initial deployment.
> It's better to mention whether adding ceph parameter apart from clientkey is
> mandatory or not.
> 
> ~~~
> parameter_defaults:
>   CephStorageCount: 3
>   CephClusterFSID: '4b5c8c0a-ff60-454b-a1b4-9747aa737d19'
>   CephMonKey: 'AQC+Ox1VmEr3BxAALZejqeHj50Nj6wJDvs96OQ==''
>   CephAdminKey: 'AQDLOh1VgEp6FRAAFzT7Zw+Y9V6JJExQAsRnRQ=='
>   CephClientKey: 'AQAfoaRXSAy/HxAAShHIViinopC2xtPW+RceQA=='
> ~~~

You mean like telling that each key has to be created with the ceph-authtool ?

I'm going through my backlog and found this BZ. I think most of the items have been implemented:

1. The image upload path is now a full path instead of the relative path.

2. All the commands should now be using "openstack stack *" instead of "heat stack-*"

3. Removed the timezone note since that was a leftover from OSP8.

4. Added the command for generating ceph keys (ceph-authtool)

5. As Sofer pointed out, the glance-registry is not user facing. (in fact, it looks like it's no longer deployed).

Vikrant has sadly left Red Hat. I'll be closing this BZ but if there's anything I've missed