tripleo-quickstart-cephadm(pacific-release)

tripleo-quickstart-ceph

git clone https://github.com/openstack/tripleo-quickstart.git
cd tripleo-quickstart
# Apply patch to use pacific repos on tripleo-quickstart
git fetch "https://review.opendev.org/openstack/tripleo-quickstart" refs/changes/43/782943/2 && git checkout FETCH_HEAD
cp config/nodes/3ctlr_2comp_3ceph.yml ~/myNodes.yml
#Apply the following patch
# Create the patch
cd ~
cat > ~/myNodes.patch <<EOF
--- tripleo-quickstart/config/nodes/3ctlr_2comp_3ceph.yml       2021-03-26 10:02:05.997970444 +0100
+++ /root/myNodes.yml      2021-04-05 11:34:26.503839431 +0200
@@ -4,11 +4,14 @@
 # nodes, plus another 12GB for the undercloud, for a total of around
 # 60GB.
 control_memory: 8192
-compute_memory: 4096
-ceph_memory: 4096
-
+control_disk: 45
+compute_memory: 5120
+compute_disk: 50
+ceph_memory: 5120
+ceph_disk: 55
 undercloud_memory: 12288
 
+
 # Giving the undercloud additional CPUs can greatly improve heat's
 # performance (and result in a shorter deploy time).
 # Using a total of 10 vcpus (1 is the default for overcloud nodes)
@@ -36,23 +39,19 @@
     flavor: compute
     virtualbmc_port: 6233
 
-  - name: compute_1
-    flavor: compute
-    virtualbmc_port: 6234
-
   - name: ceph_0
     flavor: ceph
-    virtualbmc_port: 6235
+    virtualbmc_port: 6234
 
   - name: ceph_1
     flavor: ceph
-    virtualbmc_port: 6236
+    virtualbmc_port: 6235
 
   - name: ceph_2
     flavor: ceph
-    virtualbmc_port: 6237
+    virtualbmc_port: 6236
 
-node_count: 8
+node_count: 7
 
 topology_map:
   Controller:
@@ -71,7 +70,7 @@
       - network: tenant
         subnet: tenant_subnet
   Compute:
-    scale: 2
+    scale: 1
     networks:
       - network: ctlplane
         vif: true

EOF
# Apply the patch
patch myNodes.yml myNodes.patch

#Deploy the environment
##### --extra-vars docker_registry_host="quay.io" --extra-vars container_build_id="current-tripleo"
cd tripleo-quickstart
./quickstart.sh  -n -R master-tripleo-ci --tags none -T all --nodes /root/myNodes.yml  127.0.0.2
./quickstart.sh  -n -X -R  master-tripleo-ci --tags all   -T none --extra-vars undercloud_disk=70 --nodes /root/myNodes.yml -p quickstart.yml 127.0.0.2
./quickstart.sh -R master-tripleo-ci --no-clone --tags all -I -T none --nodes /root/myNodes.yml  --extra-vars undercloud_enable_swift=true -p quickstart-extras-undercloud.yml 127.0.0.2
./quickstart.sh -R master-tripleo-ci --no-clone --tags all -I -T none  --nodes /root/myNodes.yml -p quickstart-extras-overcloud-prep.yml 127.0.0.2
./quickstart.sh -R master-tripleo-ci --no-clone --tags overcloud-scripts -I -T none  --nodes /root/myNodes.yml -p quickstart-extras-overcloud.yml 127.0.0.2 

###CEPH PART
=============
# Ensure that the ceph nodes use the /dev/vda as a root_device
[stack@undercloud ~]$ source stackrc
(undercloud) [stack@undercloud ~]$ openstack baremetal node set ceph-0 --property root_device="{ \"name\" : \"/dev/vda\" }"
(undercloud) [stack@undercloud ~]$ openstack baremetal node set ceph-1 --property root_device="{ \"name\" : \"/dev/vda\" }"
(undercloud) [stack@undercloud ~]$ openstack baremetal node set ceph-2 --property root_device="{ \"name\" : \"/dev/vda\" }"

# Create the topology where the VMs are going to act as what we want on the overcloud
# The overcloud_baremetal_deploy.yaml does not have the instances, but it has the networks so
# instances need to be added to the line. 
[stack@undercloud ~]$ vi overcloud_baremetal_deploy.yaml
- name: Controller
  count: 3
  instances:
  - hostname: overcloud-controller-0
    name: control-0 
  - hostname: overcloud-controller-1
    name: control-1
  - hostname: overcloud-controller-2
    name: control-2
- name: Compute
  count: 1
  instances:
  - hostname: overcloud-novacompute-0
    name: compute-0
- name: CephStorage
  count: 3
  instances:
  - hostname: overcloud-cephstorage-0
    name: ceph-0
  - hostname: overcloud-cephstorage-1
    name: ceph-1
  - hostname: overcloud-cephstorage-2
    name: ceph-2

# Provision the servers properly, I needed to do it until all the servers are deploy properly
### Comment that part on the overcloud-deply.sh script
## * Provision the baremetal nodes
## ::
#openstack overcloud node provision \
#   \
#  --stack overcloud  \
#  --output /home/stack/overcloud-baremetal-deployed.yaml /home/stack/overcloud_baremetal_deploy.yaml
#if [ $? -ne 0 ]; then
#    exit 1
#fi
(undercloud) [stack@undercloud ~]$ openstack overcloud network provision -y -o overcloud-networks-deployed.yaml /usr/share/openstack-tripleo-heat-templates/ci/network_data.yaml
(undercloud) [stack@undercloud ~]$ openstack overcloud network vip provision -y -o overcloud-vips-deployed.yaml  --stack overcloud /usr/share/openstack-tripleo-heat-templates/ci/vip_data.yaml
(undercloud) [stack@undercloud ~]$ openstack overcloud node provision -o overcloud-baremetal-deployed.yaml --stack overcloud  overcloud_baremetal_deploy.yaml
(undercloud) [stack@undercloud ~]$ openstack overcloud node provision --stack overcloud  --output /home/stack/overcloud-baremetal-deployed.yaml /home/stack/overcloud_baremetal_deploy.yaml

### If there is an error with some servers you can undeploy it manually and redeploy all (only the non-deployed are going to be deployed)
# To check the status
(undercloud) [stack@undercloud ~]$ metalsmith list
# To undeploy one server manually
(undercloud) [stack@undercloud ~]$ metalsmith undeploy NODEID

#Modify the ceph_tag on the containers-prepare-parameter.yaml file
# Set the ceph_tag: "v6.0.0-stable-6.0-pacific-centos-8-x86_64"

#Add the file /usr/share/openstack-tripleo-heat-templates/environments/cephadm/cephadm.yaml to the overcloud-deploy.sh script
# "-e /usr/share/openstack-tripleo-heat-templates/environments/cephadm/cephadm.yaml"

#Execute the overcloud-deploy.sh script
(undercloud) [stack@undercloud ~]$ ./overcloud-deploy.sh

Hello,
Tried this method many times but fails
Do you have any update on ceph install with controller HA?

Thanks

hi,
Most of the patches needed to deploy with ceph are already there.
Because the use of master, it depends on the error you hit.
The most common ones:

• Issues with the repos
• Some errors with the order of some components which makes to fail
  I'm pretting much using that procedure, with very little change or no changes at all.