Placement¶
This page is the starting point for Placement, the placement role: one container, placement_api
(roles/placement/defaults/main.yml:2-26), on the control nodes of the example inventory
(inventory/multinode:171-172, inventory/multinode:505-506), selected by the tag placement (site.yml:593-606).
enable_placement defaults to enable_nova or enable_zun (group_vars/all.yml:837). Its bootstrap container runs
the DB sync and the online data migrations on every deploy and reconfigure (Component dependencies
§3). The role is unchanged since 2023.1 except for the module rename
(kolla_docker → kolla_container) and the few points listed below CODE. The lists below say where the study covers
how to operate it, its impact, its dependencies and the evidence, what to check around a run and what stays open; the
service card follows them.
What changed in 2024.1¶
- Endpoints and the external frontend CODE:
placement_internal_endpointandplacement_public_endpointare built with thekolla_urlfilter (roles/placement/defaults/main.yml:92-93), and the public one usesplacement_api_public_port, which becomes 443 or 80 with the single external frontend option (group_vars/all.yml:558). The external HAProxy entry now takesexternal_fqdnand that port (roles/placement/defaults/main.yml:19-26). With the default (haproxy_single_external_frontendoff) the port stays8780(group_vars/all.yml:556). - Apache error log file name CODE: the WSGI virtual host writes its errors to
placement-api-error.log(roles/placement/templates/placement-api-wsgi.conf.j2:31); 2023.1 wrote them toplacement-api.log. The access log is stillplacement-api-access.log(roles/placement/templates/placement-api-wsgi.conf.j2:33). - fluentd now reads the placement logs CODE:
placementis influentd_input_openstack_services(roles/common/defaults/main.yml:201-202), so the global fluentd input tails/var/log/kolla/placement/*.log(roles/common/templates/conf/input/00-global.conf.j2:5); the 2023.1 list had noplacemententry. This changes the fluentd config of thecommonrole, not the placement container. - No RabbitMQ change applies CODE:
placement.confhas no messaging section (roles/placement/templates/placement.conf.j2:1-55), so the quorum-queue defaults of F17 do not reach it. - Still holds in 2024.1 CODE: the single handler, skipped when
kolla_actionisconfig(roles/placement/handlers/main.yml:2-16); the deploy flow register → clone (if) → config → check-containers → bootstrap → flush (roles/placement/tasks/deploy.yml:2-14); the play still takesserialfromkolla_serialand now also setsmax_fail_percentage, default 100 (site.yml:598-602), which never stops the play early (Controlled batches §1.3). The first 2024.1 check of a container whose unit 2023.1 wrote reports drift (F16).
Where to look¶
Operate¶
- Scenario: Reconfigure a single service §2, Reconfigure a single service §3
- Command: Command quick lookup by tag
Impact and risk¶
- What the tag executes (
placement): Tags §3, row<svc>; Tags §3.1, row<svc>API service - Risks: R8, R24
Dependencies¶
- Dependencies: Component dependencies §3
Evidence¶
- Fact sheet: fact sheet C §2
- Image side: Kolla images
§4 (row
placement-api), Kolla images §5
Before and after a run¶
- Tags §3.1, row
<svc>API service: safe on its own "yes, it is the intended unit; it still runs Keystone registration and DB sync"; batch withANSIBLE_SERIAL. Prerequisites: MariaDB, RabbitMQ, memcached, Keystone (registration), HAProxy VIP. - fact sheet C §2.8: rotating
placement_keystone_passwordneeds a reconfigure of placement and nova (all computes) and neutron; cyborg also renders it, when enabled (enable_cyborg, defaultno) CODE (roles/placement/templates/placement.conf.j2:36,roles/nova/templates/nova.conf.j2:196,roles/nova-cell/templates/nova.conf.j2:238,roles/neutron/templates/neutron.conf.j2:200,roles/cyborg/templates/cyborg.conf.j2:38;group_vars/all.yml:762). No other template underansible/rolesrenders it (grep). - R8, precautions for a HAProxy config change:
ANSIBLE_SERIAL=1for haproxy changes;--skip-tags haproxywhen you don't intend to change frontends. - R24, container stop timeout: in 2024.1 the stop goes through the container's systemd
unit, whose
ExecStopis/usr/bin/<engine> stop <name> -t <graceful_timeout>(module_utils/kolla_systemd_worker.py:32), withdocker_graceful_timeout60 s (group_vars/all.yml:139) (F3). Precautions: drain long operations before restarts; raisedocker_graceful_timeoutif needed (it reaches the unit on the next restart). - Reconfigure a single service
§3,
verification of
-t <svc> reconfigure: before,kolla-ansible -t <svc> genconfig, thenkolla_set_configs --checkin each<svc>container and a diff of/etc/kolla/haproxy/services.d/<svc>*.cfgagainst the previous version; after,openstack endpoint list --service <name>is unchanged for each Keystone service the role registers, the API answers through the VIP, the container start times show which hosts restarted, and the haproxy container's start time shows whether HAProxy restarted. For Placement, look for Apache errors inplacement-api-error.log, no longer inplacement-api.log(above).
Known limits and open points¶
- Known bugs and limitations, second row
site.yml: placement haproxy include lackswhen: enable_placement(site.yml:248-251). Status: still holds in 2024.1 CODE. - Q14, answered from the source CODE: Placement's
KOLLA_BOOTSTRAPrunsdb sync+online_data_migrationson every deploy/reconfigure (roles/placement/tasks/bootstrap_service.yml:10-12,roles/placement/tasks/deploy.yml:11,roles/placement/tasks/bootstrap.yml:38; kolladocker/placement/placement-api/extend_start.sh:5-9). The kolladocker/placementdirectory is unchanged from 2023.1 (diff of the two trees). - F15, "What the bootstrap containers run differs per image": Placement's
KOLLA_BOOTSTRAPcontainer also runsonline_data_migrations, on every deploy/reconfigure CODE. - F5, "
reconfigure=deploy": a reconfigure re-runs Keystone registration, DB creation/grants and the DB-sync bootstrap containers of most services CODE. For Placement,reconfigure.ymlis an import ofdeploy.yml(roles/placement/tasks/reconfigure.yml:2). - What is not known, row on upstream Nova behaviour when a dependency is missing: for Placement, the sources do not establish whether scheduling stops while it restarts.
- Nova deep dive §3, item 4: Placement serves the scheduler's allocation candidates and nova-compute's resource-provider updates; NOT AVAILABLE whether scheduling stops and computes log errors without it.
- fact sheet C §2.8, "Operational notes and the relation to nova", three points: consumers authenticate as
the placement user (
placement_keystone_user,group_vars/all.yml:858; e.g.roles/nova/templates/nova.conf.j2:195-196); NOT AVAILABLE (upstream nova): how scheduling and the nova-compute resource tracker periodic task behave during a placement-api outage or restart; the upgrade db sync and online data migrations run while the old placement-api is still serving. - Kolla images §6, item 6: Placement's online data migrations run on
every deploy/reconfigure, inside
bootstrap_placement, and twice on upgrade (bootstrap_placementwithKOLLA_BOOTSTRAP, then again withKOLLA_OSMafter the handler flush,roles/placement/tasks/upgrade.yml:11-34); a tag change applied withdeploytherefore runs them CODE.
Service card¶
The rows of the card are explained on the Components page.
| Item | Value |
|---|---|
| Play | site.yml:593 |
| Tags | placement |
| Play host patterns | placement-api |
| Node types in play (example inventory) | control |
| Node types running its containers | control |
| Containers | placement_api |
deploy/reconfigure flow (tasks/) |
register → clone (if) → config → check-containers → bootstrap → flush handlers |
| Task files | bootstrap.yml, bootstrap_service.yml, check-containers.yml, check.yml, clone.yml, config.yml, config_validate.yml, copy-certs.yml, deploy-containers.yml, deploy.yml, loadbalancer.yml, main.yml, precheck.yml, pull.yml, reconfigure.yml, register.yml, stop.yml, upgrade.yml |
| Templates | migrate-db.rc.j2, placement-api-wsgi.conf.j2, placement-api.json.j2, placement.conf.j2 |
| Handlers (definition order) | Restart placement-api container |
| Key variables | enable_placement |
| Role dependencies | none |
Related¶
- Reconfigure a single service — what
-t <svc> reconfigureruns for an API service such as placement, with the risks and the verification - Nova — the scheduler and nova-compute use Placement
- Nova deep dive §3 — Placement among what must be available before Nova can operate
- Neutron — one of the consumers that authenticate as the placement user
- Update container images §4 — a tag change applied with
deployruns no online data migrations, except Placement's - Component dependencies §6 — Placement is deployed before Nova
- Apply common changes — the fluentd input list, which now includes placement,
belongs to the
commonrole - Components — the other component pages