Permalink
Cannot retrieve contributors at this time
Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.
Sign up
Fetching contributors…
Cannot retrieve contributors at this time
| # This file can be used directly by 'phd', see 'build-all.sh' in this | |
| # directory for how it can be invoked. The only requirement is a list | |
| # of nodes you'd like it to modify. | |
| # | |
| # The scope of each command-block is controlled by the preceeding | |
| # 'target' line. | |
| # | |
| # - target=all | |
| # The commands are executed on evey node provided | |
| # | |
| # - target=local | |
| # The commands are executed from the node hosting phd. When not | |
| # using phd, they should be run from some other independant host | |
| # (such as the puppet master) | |
| # | |
| # - target=$PHD_ENV_nodes{N} | |
| # The commands are executed on the Nth node provided. | |
| # For example, to run on only the first node would be target=$PHD_ENV_nodes1 | |
| # | |
| # Tasks to be performed at this step include: | |
| ################################# | |
| # Scenario Requirements Section # | |
| ################################# | |
| = REQUIREMENTS = | |
| nodes: 1 | |
| = VARIABLES = | |
| PHD_VAR_deployment | |
| PHD_VAR_osp_configdir | |
| PHD_VAR_network_domain | |
| PHD_VAR_network_internal | |
| PHD_VAR_network_hosts_gateway | |
| ###################### | |
| # Deployment Scripts # | |
| ###################### | |
| = SCRIPTS = | |
| target=all | |
| .... | |
| if [ $PHD_VAR_deployment = segregated ]; then | |
| echo "We don't document managed compute nodes in a segregated environment yet" | |
| # Certainly none of the location constraints would work and the | |
| # resource-discovery options are mostly redundant | |
| exit 1 | |
| fi | |
| mkdir -p /etc/pacemaker | |
| cp $PHD_VAR_osp_configdir/pcmk-authkey /etc/pacemaker/authkey | |
| .... | |
| target=$PHD_ENV_nodes1 | |
| .... | |
| # add NovaEvacuate. It must be A/P and that is perfectly acceptable | |
| # to avoid need of a cluster wide locking | |
| pcs resource create nova-evacuate ocf:openstack:NovaEvacuate auth_url=http://vip-keystone:35357/v2.0/ username=admin password=keystonetest tenant_name=admin | |
| # without any of those services, nova-evacuate is useless | |
| # later we also add a order start on nova-compute (after -compute is defined) | |
| for i in vip-glance vip-cinder vip-neutron vip-nova vip-db vip-rabbitmq vip-keystone cinder-volume; do | |
| pcs constraint order start $i then nova-evacuate | |
| done | |
| for i in glance-api-clone neutron-metadata-agent-clone nova-conductor-clone; do | |
| pcs constraint order start $i then nova-evacuate require-all=false | |
| done | |
| # Take down the ODP control plane | |
| pcs resource disable openstack-keystone --wait=240 | |
| # Take advantage of the fact that control nodes will already be part of the cluster | |
| # At this step, we need to teach the cluster about the compute nodes | |
| # | |
| # This requires running commands on the cluster based on the names of the compute nodes | |
| controllers=$(cibadmin -Q -o nodes | grep uname | sed s/.*uname..// | awk -F\" '{print $1}') | |
| for controller in ${controllers}; do | |
| pcs property set --node ${controller} osprole=controller | |
| done | |
| # Force services to run only on nodes with osprole = controller | |
| # | |
| # Importantly it also tells Pacemaker not to even look for the services on other | |
| # nodes. This helps reduce noise and collisions with services that fill the same role | |
| # on compute nodes. | |
| stonithdevs=$(pcs stonith | awk '{print $1}') | |
| for i in $(cibadmin -Q --xpath //primitive --node-path | tr ' ' '\n' | awk -F "id='" '{print $2}' | awk -F "'" '{print $1}' | uniq); do | |
| found=0 | |
| if [ -n "$stonithdevs" ]; then | |
| for x in $stonithdevs; do | |
| if [ $x = $i ]; then | |
| found=1 | |
| fi | |
| done | |
| fi | |
| if [ $found = 0 ]; then | |
| pcs constraint location $i rule resource-discovery=exclusive score=0 osprole eq controller | |
| fi | |
| done | |
| # Now (because the compute nodes have roles assigned to them and keystone is | |
| # stopped) it is safe to define the services that will run on the compute nodes | |
| # neutron-openvswitch-agent | |
| pcs resource create neutron-openvswitch-agent-compute systemd:neutron-openvswitch-agent --clone interleave=true --disabled --force | |
| pcs constraint location neutron-openvswitch-agent-compute-clone rule resource-discovery=exclusive score=0 osprole eq compute | |
| pcs constraint order start neutron-server-clone then neutron-openvswitch-agent-compute-clone require-all=false | |
| # libvirtd | |
| pcs resource create libvirtd-compute systemd:libvirtd --clone interleave=true --disabled --force | |
| pcs constraint location libvirtd-compute-clone rule resource-discovery=exclusive score=0 osprole eq compute | |
| pcs constraint order start neutron-openvswitch-agent-compute-clone then libvirtd-compute-clone | |
| pcs constraint colocation add libvirtd-compute-clone with neutron-openvswitch-agent-compute-clone | |
| # openstack-ceilometer-compute | |
| pcs resource create ceilometer-compute systemd:openstack-ceilometer-compute --clone interleave=true --disabled --force | |
| pcs constraint location ceilometer-compute-clone rule resource-discovery=exclusive score=0 osprole eq compute | |
| pcs constraint order start ceilometer-notification-clone then ceilometer-compute-clone require-all=false | |
| pcs constraint order start libvirtd-compute-clone then ceilometer-compute-clone | |
| pcs constraint colocation add ceilometer-compute-clone with libvirtd-compute-clone | |
| # nfs mount for nova-compute shared storage | |
| pcs resource create nova-compute-fs Filesystem device="${PHD_VAR_network_internal}.1:$PHD_VAR_osp_configdir/instances" directory="/var/lib/nova/instances" fstype="nfs" options="v3" op start timeout=240 --clone interleave=true --disabled --force | |
| pcs constraint location nova-compute-fs-clone rule resource-discovery=exclusive score=0 osprole eq compute | |
| pcs constraint order start ceilometer-compute-clone then nova-compute-fs-clone | |
| pcs constraint colocation add nova-compute-fs-clone with ceilometer-compute-clone | |
| # nova-compute | |
| pcs resource create nova-compute ocf:openstack:NovaCompute auth_url=http://vip-keystone:35357/v2.0/ username=admin password=keystonetest tenant_name=admin domain=${PHD_VAR_network_domain} op start timeout=300 --clone interleave=true --disabled --force | |
| pcs constraint location nova-compute-clone rule resource-discovery=exclusive score=0 osprole eq compute | |
| pcs constraint order start nova-conductor-clone then nova-compute-clone require-all=false | |
| pcs constraint order start nova-compute-fs-clone then nova-compute-clone require-all=false | |
| pcs constraint colocation add nova-compute-clone with nova-compute-fs-clone | |
| pcs constraint order start nova-compute-clone then nova-evacuate require-all=false | |
| case ${PHD_VAR_network_hosts_gateway} in | |
| east-*) | |
| pcs stonith create fence-compute fence_apc ipaddr=east-apc login=apc passwd=apc pcmk_host_map="east-01:2;east-02:3;east-03:4;east-04:5;east-05:6;east-06:7;east-07:9;east-08:10;east-09:11;east-10:12;east-11:13;east-12:14;east-13:15;east-14:18;east-15:17;east-16:19;" --force | |
| ;; | |
| mrg-*) | |
| pcs stonith create fence-compute fence_apc_snmp ipaddr=apc-ap7941-l2h3.mgmt.lab.eng.bos.redhat.com power_wait=10 pcmk_host_map="mrg-07:10;mrg-08:12;mrg-09:14" | |
| ;; | |
| esac | |
| pcs stonith create fence-nova fence_compute auth-url=http://vip-keystone:35357/v2.0/ login=admin passwd=keystonetest tenant-name=admin domain=${PHD_VAR_network_domain} record-only=1 action=off --force | |
| # while this is set in basic.cluster, it looks like OSPd doesn't set it. | |
| pcs resource defaults resource-stickiness=INFINITY | |
| # allow compute nodes to rejoin the cluster automatically | |
| # 1m might be a bit aggressive tho | |
| pcs property set cluster-recheck-interval=1min | |
| for node in ${PHD_ENV_nodes}; do | |
| found=0 | |
| short_node=$(echo ${node} | sed s/\\..*//g) | |
| for controller in ${controllers}; do | |
| if [ ${short_node} = ${controller} ]; then | |
| found=1 | |
| fi | |
| done | |
| if [ $found = 0 ]; then | |
| # We only want to execute the following _for_ the compute nodes, not _on_ the compute nodes | |
| # Rather annoying | |
| pcs resource create ${short_node} ocf:pacemaker:remote reconnect_interval=60 op monitor interval=20 | |
| pcs property set --node ${short_node} osprole=compute | |
| pcs stonith level add 1 ${short_node} fence-compute,fence-nova | |
| fi | |
| done | |
| pcs resource enable openstack-keystone | |
| pcs resource enable neutron-openvswitch-agent-compute | |
| pcs resource enable libvirtd-compute | |
| pcs resource enable openstack-ceilometer-compute | |
| pcs resource enable nova-compute-fs | |
| pcs resource enable nova-compute | |
| # cleanup after us | |
| sleep 60 | |
| pcs resource cleanup | |
| .... |