Skip to content

Component dependencies

This page maps what depends on what, and keeps four kinds of dependency apart:

Kind Question it answers Where it lives in the code
Installation (deploy-time) dependency What must already run for the role to succeed? kolla_toolbox calls to MariaDB, Keystone and RabbitMQ in bootstrap.yml/register.yml, bootstrap containers, delegate_to targets
Configuration dependency Whose hosts, addresses, flags or passwords end up in my config files? templates/*.j2, group_vars/all.yml (e.g. rpc_transport_url)
Runtime dependency What must be reachable for the service to work? The same templates (URLs, connections); upstream OpenStack behaviour
Operational dependency Which Kolla action on X changes or restarts Y? Handlers, global enable_* flags, common, the loadbalancer play, loop-notify (see Operations)

Kolla-Ansible never declares inter-service dependencies explicitly. As in 2023.1, no role under ansible/roles/ has a meta directory, and common is its own play (site.yml:88-103) CODE. Every dependency below is implicit: play order in site.yml, a variable, a template, or a delegate_to.

Note

Key answers

  • The database and etcd are reached through the internal VIP and HAProxy; RabbitMQ and memcached hosts are listed one by one in the configs of the services that use them. Adding or removing a RabbitMQ node therefore changes the config of every messaging service, including nova-compute and the neutron agents on every compute; 27 roles loop over the memcached group (33 in 2023.1) CODE (§1).
  • Changed in 2024.1: RabbitMQ quorum queues are on by default, so the same messaging configs carry rabbit_quorum_queue = true (F17); keystone_admin_url no longer exists; Octavia's jobboard (on by default with the amphora driver) needs Redis, which is off by default CODE (§1, §3).
  • A compute host's configuration depends on global state (enabled services, the membership of the RabbitMQ and memcached groups, the internal VIP, service passwords), not only on the compute host itself. That is why changes that look small trigger fleet-wide restarts CODE (§5).
  • What lands on a compute host, container by container, with the global config source of each (§5).
  • Per-service dependencies: control-plane services with their use of the DB, RabbitMQ, memcached and Keystone and their deploy-time actions (§3); infrastructure components (§2); networking containers and their placement (§4).
  • L3 agent: with enable_neutron_agent_ha: yes, the L3 agent wrapper runs neutron-netns-cleanup --force --agent-type l3 before the agent starts, on every restart or recreate of neutron_l3_agent on that host CODE; what that command deletes, and the DHCP and router outage per agent restart, are NOT AVAILABLE (§4).
  • Deployment order mostly respects runtime dependencies, with caveats: Nova is deployed before Open vSwitch, OVN and Neutron, and nothing enforces a dependency when you use tags CODE (§6).

A compact card per service (role, play, tags, task flow, templates, handlers, key variables, node types) is in Components. Per-role detail with line references is in the fact sheets (A, B, C, D, E). Nova is covered in depth in Nova deep dive.


1. The shared substrate: what every OpenStack service uses

Shared component How services reach it Evidence Consequence
MariaDB / Galera database_address: "{{ kolla_internal_fqdn }}", i.e. through the internal VIP and HAProxy ansible/group_vars/all.yml:49, :83 Every DB connection depends on HAProxy and keepalived as well as on Galera. HAProxy sends DB traffic to one active backend; the others are backup (roles/mariadb/defaults/main.yml:29, :64). With enable_proxysql: yes (default no, group_vars/all.yml:839) the MariaDB load balancer is ProxySQL instead (mariadb_loadbalancer, group_vars/all.yml:481) (fact sheet B §1)
RabbitMQ rpc_transport_url / notify_transport_url loop over every host of groups['rabbitmq'] (om_rpc_group, om_notify_group). Clients use their own failover, not the VIP ansible/group_vars/all.yml:244, :248, :254, :258; nova-cell builds its own URLs from the same group by default (roles/nova-cell/defaults/main.yml:191, :200, :206-207) Adding or removing a RabbitMQ node changes the config of every messaging service, including nova-compute and the neutron agents on every compute (Change control-plane membership §2). Changed in 2024.1: om_enable_rabbitmq_quorum_queues defaults to true (group_vars/all.yml:268), rendered as rabbit_quorum_queue = true in the [oslo_messaging_rabbit] section of these configs (e.g. roles/nova-cell/templates/nova.conf.j2:207-216) (F17)
Memcached Templates loop {% for host in groups['memcached'] %} for keystone_authtoken / [cache] e.g. roles/ceilometer/templates/ceilometer.conf.j2:69; the templates of 27 roles contain this loop (33 in 2023.1: freezer, murano, sahara, senlin, solum and vitrage are gone). A 28th, telegraf, only tests whether the host itself is in the group (roles/telegraf/templates/telegraf.conf.j2:54-57) Membership change = config change in 27 services (Memcached). NOT AVAILABLE Whether a memcached restart empties the token caches and the Horizon sessions kept there (fact sheet A §3.8, Kolla images §6 item 14)
Keystone www_authenticate_uri/auth_url → internal VIP endpoint (keystone_internal_url, group_vars/all.yml:960); service-ks-register at deploy time roles/service-ks-register/tasks/main.yml (31 roles use it; 37 in 2023.1) Hard deploy-time dependency: registration runs run_once (roles/service-ks-register/tasks/main.yml:126) with the auth each role passes, e.g. openstack_nova_auth (roles/nova/tasks/register.yml:5), which is openstack_auth (roles/nova/defaults/main.yml:215), whose auth_url is keystone_internal_url (group_vars/all.yml:990-991). No longer holds in 2024.1: keystone_admin_url is gone from group_vars/all.yml (2023.1 defined it as the internal URL). Runtime: the services' [keystone_authtoken] sections point keystonemiddleware at Keystone, with memcached as token cache (e.g. roles/nova/templates/nova.conf.j2:112-127); NOT AVAILABLE whether API requests fail while Keystone is down
etcd (optional, enable_etcd: "no", group_vars/all.yml:764) Through the internal VIP: backend_url = etcd3+…://{{ kolla_internal_fqdn }}:{{ etcd_client_port }}; the etcd role defines an internal HAProxy frontend roles/cinder/templates/cinder.conf.j2:271-281, roles/masakari/templates/masakari.conf.j2:97, roles/ironic/templates/ironic-inspector.conf.j2:103; frontend roles/etcd/defaults/main.yml:11-17; Redis wins over etcd when both are on (cinder_coordination_backend, group_vars/all.yml:1047; masakari :491, ironic :1307) etcd membership is not embedded in service configs: outside the etcd role, only Prometheus's config loops over groups["etcd"] (roles/prometheus/templates/prometheus.yml.j2:214). Changed in 2024.1: the etcd play has no serial (site.yml:470-481), and its handlers restart non-leaders in % 4 waves, then the leader (roles/etcd/handlers/main.yml:40-65); on upgrade the order across hosts is NOT AVAILABLE (:28-36) (Change control-plane membership)
HAProxy + keepalived Every internal/admin/public endpoint and the DB address resolve to the VIPs roles/*/defaults/main.yml haproxy: dicts → each role's tasks/loadbalancer.yml → roles/loadbalancer-config → roles/haproxy-config (and roles/proxysql-config with enable_proxysql, roles/loadbalancer-config/tasks/main.yml:1-8) A single point of path failure for APIs, DB and etcd. HAProxy is restarted, never reloaded: the handlers stop and start it on the backup hosts first, then on the host that holds the VIP (roles/loadbalancer/handlers/main.yml:36-174, :179-305) (fact sheet A §1.8). New in 2024.1: a Restart haproxy-ssh container handler (:331-345) for the haproxy_ssh container, enabled only with enable_letsencrypt (roles/loadbalancer/defaults/main.yml:29-36) (HAProxy and keepalived)
fluentd / cron / kolla_toolbox (common) Deployed on the hosts of the groups fluentd, kolla-toolbox and cron (children of common: control, network, compute, storage, monitoring) roles/common/defaults/main.yml:5, :14, :26; site.yml:88-103; inventory/multinode:49-54, :203-213 kolla_toolbox is a deploy-time dependency of every role that touches MySQL, Keystone, RabbitMQ or the OVS database (kolla_toolbox module needs it running). fluentd is runtime-optional (logging only) (Common)
flowchart LR
  subgraph Substrate["Shared substrate (runtime)"]
    VIP["keepalived VIP"] --> HAP["HAProxy (haproxy hosts)"]
    HAP --> DB[("MariaDB/Galera<br/>1 active backend")]
    HAP -.optional.-> ETCD[("etcd<br/>(if enable_etcd)")]
    MQ[["RabbitMQ cluster<br/>(direct, all nodes listed)"]]
    MC[("Memcached<br/>(direct, all nodes listed)")]
  end
  KS["Keystone API"] -->|SQL via VIP| HAP
  KS -->|token cache| MC
  SVC["Any OpenStack API / agent"] -->|SQL via VIP| HAP
  SVC -->|RPC + notifications| MQ
  SVC -->|keystonemiddleware cache| MC
  SVC -->|token validation, via VIP| HAP
  HAP -->|"/v3"| KS

2. Infrastructure components

Component Role / play Hosts (example inventory) Deploy-time needs Runtime needs Who depends on it
Keepalived loadbalancer role, site.yml:105-120 loadbalancer → network (inventory/multinode:84-85) ip_vs module, sysctl ip_nonlocal_bind, now set through the sysctl role (roles/loadbalancer/tasks/config-host.yml:8-25) VRRP between loadbalancer hosts (multicast default, roles/loadbalancer/defaults/main.yml:62; router id 51, group_vars/all.yml:224) Everything addressed by VIP
HAProxy loadbalancer role + 39 loadbalancer.yml includes (site.yml:105-327; 44 in 2023.1) loadbalancer facts of all backend hosts (backend lists use hostvars[host].ansible_facts.hostname, roles/haproxy-config/templates/haproxy_single_service_split.cfg.j2:102) backends All APIs, MariaDB (DB address = VIP), etcd
Let's Encrypt (new in 2024.1) letsencrypt play, right after the loadbalancer play (site.yml:328-340) letsencrypt → loadbalancer (inventory/multinode:192-193) only if enable_letsencrypt (default no, group_vars/all.yml:796) – HAProxy certificates (haproxy_ssh)
MariaDB/Galera mariadb role, mariadb.yml:3-16 (no serial; restarts in separate plays with serial: "33%", mariadb.yml:18-52) mariadb → control kolla_toolbox; Galera bootstrap logic (fact sheet B §1) HAProxy (clustercheck on port 4569, group_vars/all.yml:471) Every service with a DB
RabbitMQ rabbitmq role, rabbitmq.yml:3-28 (no serial; restarts with serial: "33%", rabbitmq.yml:30-49) rabbitmq → control erlang cookie, /etc/hosts (from bootstrap-servers, ansible-collection-kolla roles/baremetal/tasks/main.yml:3, role etc_hosts) – Every service with RPC/notifications (RabbitMQ)
Memcached memcached role, site.yml:402-414 memcached → control – – Keystone, all keystonemiddleware users, Horizon sessions
etcd etcd role, site.yml:470-481 (no serial in 2024.1) etcd → control (inventory/multinode:62-63) only if enable_etcd – Cinder, Masakari, ironic-inspector coordination when chosen (§1)
Redis redis role, site.yml:384-398 redis → control (inventory/multinode:183-184) only if enable_redis (default no, group_vars/all.yml:840) – Coordination backends when enabled; Octavia jobboard (§3)
Fluentd common play, site.yml:88-103 fluentd → common (all node types) the fluentd image, pulled and inspected even by a config-only run (roles/common/tasks/config.yml:42-91, Common) (optional) Elasticsearch or OpenSearch outputs (roles/common/tasks/config.yml:140-203) Logging only. HAProxy syslogs to fluentd UDP (roles/loadbalancer/templates/haproxy/haproxy_main.cfg.j2:7, group_vars/all.yml:630)
Open vSwitch openvswitch role, site.yml:611-624 openvswitch → network, compute, manila-share (inventory/multinode:105-108) kernel module openvswitch (roles/openvswitch/tasks/config-host.yml:2-7); neutron_bridge_name/neutron_external_interface on network nodes (and computes if computes_need_external_bridge) – neutron-openvswitch-agent, OVN controller, DHCP/L3 agents
OVN (if neutron_plugin_agent: ovn) ovn-controller role, site.yml:641, and ovn-db role, site.yml:655 ovn-database → control; ovn-controller → compute, network (inventory/multinode:593-604) OVS running (ovn-controller's setup creates br-int and writes OVS external_ids through the openvswitch_bridge and openvswitch_db modules in kolla_toolbox, roles/ovn-controller/tasks/setup-ovs.yml:2-40) NB/SB DB cluster (Raft) neutron-server (ML2/OVN), neutron-ovn-metadata-agent
Chrony none in 2024.1: no roles/chrony, no chrony play in site.yml, no chrony group in inventory/multinode (Chrony) – (host NTP daemon, outside Kolla-Ansible) kolla-ansible prechecks checks that a host NTP daemon is active and the clock synchronised (roles/prechecks/tasks/timesync_checks.yml:1-47) the host's NTP daemon NOT AVAILABLE in the code: which services need synchronised clocks
iscsid / tgtd iscsi role, site.yml:439-452 iscsid → compute, storage, ironic; tgtd → storage (inventory/multinode:305-311) only if enable_iscsid (cinder with the iSCSI backend, group_vars/all.yml:794) – nova-compute (volume attach), cinder-volume (LVM) (iSCSI)

2.1 Infrastructure dependency graph (runtime)

flowchart TB
  KA["keepalived"] -->|VIP| HAP["HAProxy"]
  HAP -->|"tcp 3306 (1 active)"| DB[("MariaDB/Galera")]
  HAP -->|HTTP| KS["Keystone"]
  HAP -.->|"2379 (if enabled)"| ETCD[("etcd")]
  KS --> MC[("Memcached")]
  KS -->|"SQL via VIP"| HAP
  OVS["openvswitch_db + vswitchd"] --> OVSAG["neutron-openvswitch-agent"]
  OVS --> OVNC["ovn-controller"]
  OVNC --> SB[("OVN SB DB")]
  NORTHD["ovn-northd"] --> NB[("OVN NB DB")] & SB
  FL["fluentd (all hosts)"] -.optional.-> ES[("Elasticsearch / OpenSearch")]
  TB["kolla_toolbox (all hosts)"] -.deploy-time only.-> DB & KS & MQ[["RabbitMQ"]]

3. Control-plane services

Legend: DB = MariaDB through the VIP, MQ = RabbitMQ (all nodes listed), MC = memcached, KS = Keystone. A ✔ means the role's config templates (roles/<role>/templates/*.j2) reference it CODE. Every role below except Swift runs its deploy steps on reconfigure too: its reconfigure.yml imports deploy.yml (e.g. roles/keystone/tasks/reconfigure.yml:2) CODE. What each bootstrap container runs inside the image is listed in Kolla images §4.

Service DB MQ MC KS Other services referenced in config Deploy-time actions Notes (fact sheet)
Keystone ✔ ✔ (roles/keystone/templates/keystone.conf.j2:4, :54) ✔ – – bootstrap container (db sync) every deploy/reconfigure, no condition (roles/keystone/tasks/bootstrap_service.yml:33-50); fernet keys pushed from the first host where keystone_fernet runs (roles/keystone/tasks/distribute_fernet.yml:12-18); admin bootstrap through kolla_keystone_bootstrap (roles/keystone/tasks/register.yml:2-11) B §3.6–3.10, Keystone
Glance ✔ ✔ ✔ ✔ Cinder (store), Ceph, Swift, S3 (new store, glance_backend_s3, default no, group_vars/all.yml:1004), Barbican db sync + db_load_metadefs on every deploy/reconfigure (kolla docker/glance/glance-api/extend_start.sh:5-9); toggles global MariaDB log_bin_trust_function_creators (roles/glance/tasks/bootstrap_service.yml:2-12, :39-49) C §1. With the file backend only groups['glance-api'][0] runs glance (group_vars/all.yml:1011)
Placement ✔ – ✔ ✔ (Nova credentials in migration helper) db sync and online data migrations on every deploy/reconfigure (one KOLLA_BOOTSTRAP container, kolla docker/placement/placement-api/extend_start.sh:5-9) C §2. The placement templates have no transport_url (the 2023.1 table showed "✔ (unused)")
Nova ✔ ✔ ✔ (API side only; nova-cell's nova.conf has no memcached list) ✔ Glance, Neutron, Placement, Cinder, Barbican, Ceph API/cell0/cell DB sync on every deploy/reconfigure; cell create/update; host discovery Nova deep dive
Neutron ✔ (server) ✔ ✔ ✔ Nova (notifications to Nova), Placement, OVN DB, user and grants, and the bootstrap_neutron db sync container, on every deploy/reconfigure (roles/neutron/tasks/bootstrap.yml:38, bootstrap_service.yml:2-21, no condition; neutron_enable_rolling_upgrade only selects the upgrade path, roles/neutron/tasks/upgrade.yml:4-8) §4 below
Cinder ✔ ✔ ✔ ✔ Glance, Nova (uses nova credentials), Barbican, Swift (backup), Ceph; etcd through the internal VIP when cinder_coordination_backend is etcd, or Redis (all hosts listed in redis_connection_string, group_vars/all.yml:893) when it is redis (roles/cinder/templates/cinder.conf.j2:271-281, group_vars/all.yml:1047) db sync on every deploy/reconfigure (roles/cinder/tasks/deploy.yml:11) C §3
Horizon only if horizon_backend_database (default false, group_vars/all.yml:1141) – ✔ (sessions, roles/horizon/templates/_9998-kolla-settings.py.j2:19-23) ✔ every enabled API (internal endpoint type, _9998-kolla-settings.py.j2:131) no Keystone registration C §4. Changed in 2024.1: the settings template is _9998-kolla-settings.py.j2 (local_settings.j2 is gone)
Heat ✔ ✔ ✔ ✔ every API used by templates (internal); instances need the external CFN endpoint db sync on every deploy/reconfigure; heat domain admin setup only when heat_user_domain does not exist yet (kolla docker/heat/heat-api/extend_start.sh:8-14): a changed heat_domain_admin_password reaches heat.conf (roles/heat/templates/heat.conf.j2:12) but is never applied in Keystone C §5
Barbican ✔ ✔ ✔ ✔ – db sync on every deploy/reconfigure (roles/barbican/tasks/deploy.yml:11); barbican_crypto_key from passwords.yml D §1
Octavia ✔ (two databases: main and persistence, roles/octavia/tasks/bootstrap.yml:2-72) ✔ ✔ ✔ Nova, Neutron, Glance (amphora image, flavor, network, secgroup); Redis for the jobboard (every host of groups['redis'], roles/octavia/templates/octavia.conf.j2:177-187) db sync; with octavia_auto_configure (true when the amphora provider is listed, the default; group_vars/all.yml:1323, roles/octavia/defaults/main.yml:386) prepare.yml creates the amphora flavor, keypair, security groups and management network, subnet and router (roles/octavia/tasks/deploy.yml:4-5, prepare.yml:2-161); the amphora image stays manual. New in 2024.1: enable_octavia_jobboard is on whenever Octavia uses the amphora provider (group_vars/all.yml:832) and the precheck fails unless enable_redis is set (roles/octavia/tasks/precheck.yml:67-73; enable_redis defaults to no, group_vars/all.yml:840) D §2
Swift – ✔ (notifications, roles/swift/templates/proxy-server.conf.j2:62) ✔ ✔ – rings are operator-built; old-style deploy flow: register, config, start, check, with no check-containers step (roles/swift/tasks/deploy.yml:2-10); its reconfigure.yml has its own flow D §3
Manila ✔ ✔ ✔ ✔ Glance, Neutron, Nova, Cinder, Ceph (CephFS) db sync; manila-share hosts also get the Open vSwitch agent with the generic backend (roles/neutron/defaults/main.yml:36) D §4

3.1 Control-plane runtime graph

flowchart LR
  USER(("API clients<br/>Horizon")) --> HAP["HAProxy VIP"]
  HAP --> KS["Keystone"] & GL["Glance API"] & PL["Placement API"] & NA["Nova API"] & NS["Neutron server"] & CA["Cinder API"] & HA["Heat API/CFN"] & BA["Barbican API"] & OA["Octavia API"] & SW["Swift proxy"] & MA["Manila API"]
  NA -->|RPC| NSC["nova-scheduler"] --> PL
  NA --> NCOND["nova-conductor"]
  NS -->|"port events (os-server-external-events)"| NA
  NA -->|ports| NS
  NA -->|images| GL
  NA -->|volumes| CA
  CA -->|images| GL
  CA -.->|"nova creds: volume events"| NA
  GL -.->|store| CA
  GL -.->|store| SW
  HA -->|"orchestrates (internal endpoints)"| NA & NS & CA & GL
  OA -->|amphorae| NA & NS & GL
  OA -.->|"jobboard"| RD[("Redis")]
  MA -->|share servers| NA & NS & CA
  NA & NS & CA & GL & PL & HA & BA & OA & MA -->|tokens| KS

The dotted edges are configuration couplings that are easy to overlook:

  • Cinder → Nova with Nova's own credentials (roles/cinder/templates/cinder.conf.j2:109-119).
  • Glance → Cinder or Swift as image stores. Enabling Swift changes Glance's default backend: glance_backend_swift follows enable_swift (group_vars/all.yml:1006), and glance_default_backend picks swift unless VMware or Ceph is set (roles/glance/defaults/main.yml:247) (fact sheet C §1.8).
  • Nova and Neutron → Placement with the placement user (roles/nova-cell/templates/nova.conf.j2:234-244).
  • Octavia → Redis, new in 2024.1: the jobboard lists every Redis host, and it is on by default with the amphora provider (roles/octavia/templates/octavia.conf.j2:177-187, group_vars/all.yml:832).

4. Networking components (ML2/OVS default, ML2/OVN optional)

Container Placement rule (host_in_groups) Needs Data-plane role
openvswitch_db, openvswitch_vswitchd compute, manila-share (generic backend), dhcp/l3/metadata agent hosts (roles/openvswitch/defaults/main.yml:2-34) kernel module, /run/openvswitch Forwards all VM traffic
neutron_openvswitch_agent compute, manila-share (generic), dhcp/l3/metadata hosts (roles/neutron/defaults/main.yml:26-46; with enable_nova_fake only the dhcp/l3/metadata hosts) OVS, RabbitMQ, (neutron-server via RPC) Programs OVS flows; NOT AVAILABLE whether and how it resyncs them after a restart (Kolla images §5)
neutron_l3_agent neutron-l3-agent; plus every compute when enable_neutron_dvr (roles/neutron/defaults/main.yml:85-89) OVS, /run/netns Routers (namespaces, keepalived for HA routers)
neutron_dhcp_agent neutron-dhcp-agent OVS dnsmasq per network
neutron_metadata_agent neutron-metadata-agent; plus computes with DVR and ML2/OVS (roles/neutron/defaults/main.yml:122-126) nova-api metadata via VIP (nova_metadata) metadata proxy
neutron_server neutron-server DB, MQ, KS, Nova, Placement control plane only
ovn_controller (OVN) ovn-controller (compute + network; role ovn-controller) OVS, SB DB Programs OVS from SB DB
neutron_ovn_metadata_agent (OVN) neutron-ovn-metadata-agent → compute, network (inventory/multinode:264-266; roles/neutron/defaults/main.yml:135) OVS, SB DB metadata per chassis
flowchart TB
  subgraph Compute["Compute host (ML2/OVS)"]
    NC["nova_compute"] -->|"vif plug (os-vif → ovsdb)"| OVSDB[("openvswitch_db")]
    NL["nova_libvirt / qemu"] -->|tap devices| VS["openvswitch_vswitchd"]
    OA["neutron_openvswitch_agent"] -->|flows| VS
    OA --> OVSDB
    VS --- OVSDB
  end
  subgraph Network["Network host"]
    L3["neutron_l3_agent<br/>(router namespaces, keepalived)"] --> VS2["openvswitch_vswitchd"]
    DHCP["neutron_dhcp_agent (dnsmasq)"] --> VS2
    MD["neutron_metadata_agent"]
    OA2["neutron_openvswitch_agent"] --> VS2
  end
  subgraph Control["Controller"]
    NS["neutron_server"]
  end
  OA & OA2 & L3 & DHCP & MD -->|RPC| MQ[["RabbitMQ"]]
  NS --> MQ
  NS -->|"port bound/unbound events"| NAPI["nova-api"]
  NC -->|"port create/update"| NS
  MD -->|"proxy to nova metadata (VIP)"| NAPI

Runtime facts CODE:

  • The L3 agent wrapper runs a namespace cleanup on start. With enable_neutron_agent_ha: yes (default no, group_vars/all.yml:816), its start wrapper runs neutron-netns-cleanup --force --agent-type l3 before starting the agent (roles/neutron/templates/neutron-l3-agent-wrapper.sh.j2:5-14; new in 2024.1, it also reads fwaas_driver.ini when enable_neutron_fwaas is set, :10-12), on every restart or recreate of neutron_l3_agent on such a host (the wrapper's comment: "Remove all l3 related netns"). NOT AVAILABLE What the command deletes: it is neutron code, which is not in these sources (F12); whether and when neutron-l3-agent rebuilds the namespaces (Risk register R6).
  • Agent containers do not share the host PID namespace. They have no pid_mode (roles/neutron/defaults/main.yml). Processes they spawn (dnsmasq, keepalived, haproxy for metadata) live in the container's PID namespace. NOT AVAILABLE Whether they die with the container and are respawned by the agent on start. For the L3 agent, its wrapper starts neutron-l3-agent without exec, and dumb-init runs with --single-child (kolla docker/base/Dockerfile.j2:370; roles/neutron/templates/neutron-l3-agent-wrapper.sh.j2:16-20) CODE. NOT AVAILABLE Whether the SIGTERM of the stop then reaches only bash and not the agent, and so whether the agent gets a graceful shutdown: that depends on how dumb-init forwards signals, which these sources do not show (Kolla images §2.6, §5.3). NOT AVAILABLE The DHCP and router outage per agent restart.
  • The OVS bridge layout is set in a post-config step, now through kolla_toolbox modules. On every deploy/reconfigure (roles/openvswitch/tasks/deploy.yml:11; reconfigure.yml imports deploy.yml), on network hosts (and computes with computes_need_external_bridge), the bridges are ensured by the openvswitch_bridge module (fail_mode: standalone) and their ports by openvswitch_port (roles/openvswitch/tasks/post-config.yml:25-53). Changed in 2024.1: the image script kolla_ensure_openvswitch_configured is no longer called. The same file first sets external_ids:system-id, the new external_ids:hostname (the host's FQDN, roles/openvswitch/defaults/main.yml:106) and other_config:hw-offload; a change there notifies Restart openvswitch-vswitchd container (post-config.yml:3-23) CODE (Open vSwitch).

5. Compute-node dependency summary

Everything that lands on a compute host in a default 2024.1 deployment (enable_* defaults in ansible/group_vars/all.yml:718-867, placement from each role's <role>_services):

Container Role Config source that is global (changes fleet-wide)
fluentd, cron, kolla_toolbox common td-agent.conf and logrotate.conf built from all enable_* flags (roles/common/tasks/config.yml:140-270)
chrony none in 2024.1 no chrony container or chrony.conf: there is no roles/chrony and no chrony play (Chrony); time sync is the host's NTP daemon
iscsid iscsi (if cinder with the iSCSI backend, group_vars/all.yml:794) –
nova_libvirt, nova_ssh, nova_compute nova-cell nova.conf: transport_url (all RabbitMQ hosts, roles/nova-cell/templates/nova.conf.j2:37, :184), quorum-queue flags (:207-216), enable_ceilometer/watcher/cinder/osprofiler/barbican… sections, Glance, Neutron and Placement endpoints, service passwords
openvswitch_db, openvswitch_vswitchd openvswitch – (bridge mappings only if computes_need_external_bridge)
neutron_openvswitch_agent neutron neutron.conf (all RabbitMQ hosts, memcached list, roles/neutron/templates/neutron.conf.j2:123) and openvswitch_agent.ini (merged from the role template, the operator's openvswitch_agent.ini and the per-host <inventory_hostname>/openvswitch_agent.ini; no ml2_conf.ini override, roles/neutron/tasks/config.yml:198-213)

Optional compute-side containers come from ceilometer, collectd, telegraf, prometheus exporters, masakari, zun, kuryr and cyborg (skydive and monasca have no role in 2024.1) (fact sheet D §5). Several of them embed host lists from other groups in their config: memcached (ceilometer, masakari, zun, cyborg), RabbitMQ through the transport URLs, and for telegraf the influxdb group among others (grep of groups[...] in roles/<role>/templates) CODE. No compute-side template lists the etcd hosts (see §1). The full list of templates that loop over groups is in Generated inventory §4.

Rule of thumb CODE: a compute host's configuration depends on global state (enabled services, the membership of the RabbitMQ and memcached groups, the internal VIP, service passwords, and in 2024.1 the RabbitMQ queue-type defaults), not only on the compute host itself. That is why changes that look small trigger fleet-wide restarts (Compute scaling, F6, Operations). Changed from 2023.1: the etcd group no longer belongs in this list, because no service template outside Prometheus embeds it.


6. Deployment order vs runtime dependency

Deployment order is the play order in site.yml (Execution flow). It mostly respects runtime dependencies, with these caveats CODE:

Observation Evidence Effect
HAProxy is configured for all services before any of them exists site.yml:105-327 (the loadbalancer play, right after common), then Let's Encrypt (:328-340) The first deploy puts frontends in place for backends not yet started. Each backend server is health-checked (check inter 2000 rise 2 fall 5, roles/haproxy-config/defaults/main.yml:14); how HAProxy treats a backend that is not up yet is NOT AVAILABLE
Redis, MariaDB, memcached, RabbitMQ and etcd come before Keystone site.yml:384 (redis), :400 (mariadb.yml), :402 (memcached), :468 (rabbitmq.yml), :470 (etcd), :483 (keystone) The substrate exists when the first service registers and syncs its DB
Placement is deployed before Nova; Glance and Cinder too site.yml:544-608 (glance :544, cinder :576, placement :593; nova.yml at :609) Nova bootstrap can use Placement and Glance at once
Nova is deployed before Open vSwitch, OVN and Neutron site.yml:609 (nova.yml) vs :611-694 On a first deploy nova-compute starts before the OVS agent exists, and Kolla waits for its registration in the nova-cell play, before the OVS and Neutron plays run (roles/nova-cell/tasks/deploy.yml:19-37, wait_discover_computes.yml:7-20) (F13). NOT AVAILABLE Whether instances can be scheduled there before those plays finish (Execution flow §2.4)
Heat, Horizon and Octavia come after Nova and Neutron site.yml:724-997 (heat :724, horizon :740, octavia :981) –
Nothing enforces a dependency when you use tags --tags filters plays -t nova on a fresh cloud reaches Keystone registration (service-ks-register, against keystone_internal_url) with no Keystone deployed. On an existing cloud it assumes MariaDB, RabbitMQ and Keystone are healthy. With -t octavia and the 2024.1 defaults, the Octavia config lists the Redis hosts whether or not Redis was deployed
Bootstrap and registration tasks target groups[...][0] e.g. roles/nova-cell/tasks/bootstrap.yml:15, :34; roles/keystone/tasks/bootstrap.yml:14, :34 The first host of a group is a hidden dependency: it must be reachable and have kolla_toolbox running

To remember

Lumo: if you keep only 5 things from this page, keep these.

  • The database and etcd are reached through the internal VIP and HAProxy; RabbitMQ and memcached hosts are listed one by one in the configs of the services that use them (27 roles loop over groups['memcached']). CODE
  • Adding or removing a RabbitMQ node changes the config of every messaging service, including nova-compute and the neutron agents on every compute; in 2024.1 the same configs also carry rabbit_quorum_queue = true by default. CODE
  • A compute host's configuration depends on global state, not only on the host itself: that is why changes that look small trigger fleet-wide restarts. CODE
  • HAProxy is restarted, never reloaded: on the backup hosts first, then on the host that holds the VIP. CODE
  • Nova is deployed before Open vSwitch, OVN and Neutron, and nothing enforces a dependency when you use tags. CODE

Check yourself

Lumo: 4 quick questions to check the main ideas of this page. Choose an answer to see the explanation and the section that gives the evidence. Your score counts your first choices and stays in this browser only.

1. How do OpenStack services reach RabbitMQ in 2024.1?
Show the answer

Answer: Every host of groups['rabbitmq'] is listed directly in their configs. rpc_transport_url / notify_transport_url loop over every host of groups['rabbitmq']; clients use their own failover, not the VIP. Adding or removing a RabbitMQ node changes the config of every messaging service, including nova-compute and the neutron agents on every compute. Read the section.

2. How does Cinder reach etcd when cinder_coordination_backend is etcd?
Show the answer

Answer: Through the internal VIP (kolla_internal_fqdn) and the etcd HAProxy frontend. The [coordination] backend_url points at kolla_internal_fqdn:etcd_client_port, and the etcd role defines an internal HAProxy frontend. No service template except Prometheus's loops over groups['etcd'], so an etcd membership change does not change compute configs. Read the section.

3. How is HAProxy restarted when its configuration changes?
Show the answer

Answer: Restarted, never reloaded: on the backup hosts first, then on the host that holds the VIP. HAProxy is restarted, never reloaded: the handlers stop and start it on the backup hosts first, then on the host that holds the VIP. Read the section.

4. Which statement about the deployment order in site.yml is true?
Show the answer

Answer: Nova is deployed before Open vSwitch, OVN and Neutron. Nova is deployed before Open vSwitch, OVN and Neutron CODE: on a first deploy nova-compute starts before the OVS agent exists. Glance, Cinder and Placement are deployed before Nova. Read the section.