CHG-0004: Omada Controller Upgrade

CHG-0004: Omada Controller Upgrade #

Date2026-09-10
Change typeUpgrade
ClassificationStructural — a controller outage, but nothing on site depended on the controller
StatusComplete 2026-09-10. Re-scoped 2026-09-11: see Scope change.
Window2026-09-10, 05:16–05:30 local
Sitemobile
SystemsOmada controller on dv00bld001p01
AutomationController pinned by inventory #24 and builder #13; runbook in docs #30
RiskA newer controller’s database cannot be opened by an older one
Related changesCHG-0005 (the AP) and CHG-0006 (the switch), both split out of this record
Related incidentsNone
Related runbooksOmada Controller Upgrade; Omada Controller Recovery

Summary #

The Omada controller ran 6.1. This change upgraded it in place to 6.3.0.45, the current release, and staged 6.2.14.11 as a rehearsed fallback. The aim was to bring the controller current before it is used to manage the site’s Omada devices. The switch’s current firmware recommends controller 6.2.0.

Scope change #

This record opened as Omada Controller and Network Firmware Upgrade, with three phases: the controller, the switch’s firmware and the AP’s firmware.

On 2026-09-11 it was split, so that the AP could be fixed without touching the switch:

WasNow
Phase 1 — controllerThis record, complete
Phase 3 — AP firmwareCHG-0005, which also adopts the AP and has the controller provision its SSIDs
Phase 2 — switch firmwareCHG-0006, a future change

The phases’ plans, baselines, risks and undo moved to those records intact.

Goal #

  • The controller runs 6.3.0.45, with its data intact and the a_autoprov automation login working.
  • 6.2.14.11 is staged and rehearsed as the fallback, and the pre-upgrade data snapshot is kept.
  • Inventory pins the controller build.

Scope #

In scope: the controller. Out of scope: device firmware, and adoption — see Scope change.

Risk and impact #

RiskGuard
A newer controller’s database cannot be opened by an older oneA clean snapshot was taken first; 6.2.14.11 was rehearsed on a copy of it
The controller upgrades device firmware on its ownautoUpgrade is off at site level (checked 2026-09-10)

Procedure #

Followed Omada Controller Upgrade: stage the image while the controller runs; stop it cleanly; snapshot the data; recreate the container on the new image; verify; pin inventory. Then 6.2.14.11 was staged as the fallback and rehearsed.

Verification #

The same omadacId and configured: true after the upgrade; Database upgraded in the server log; a_autoprov logging in over the API and listing the site.

Undo #

Omada Controller Recovery: restore the pre-upgrade snapshot, and start it on 6.2.14.11, or on 6.1 as the last resort. Anything configured on 6.3 since the snapshot is lost.


Outcome #

Times are local, from file, log and commit timestamps.

TimeWhat happened
05:166.3.0.45 pulled while 6.1.0.19 kept running, and saved to the artifact server
05:17Controller stopped. mongod.log ended with mongod shutdown complete. Data snapshot taken, 138 MB, with its sha256 recorded alongside.
05:18Container recreated on 6.3.0.45. Upgrading the database at 05:18:54, Database upgraded a second later, Omada Network Application started at 05:18:59.
shortly afterVerified: the same omadacId, still configured; a_autoprov logged in over the API and listed the dvntm site
05:266.2.14.11 pulled and saved as the fallback
05:27–05:30Fallback rehearsed. 6.2.14.11 was started on a copy of the snapshot, in an isolated container on a loopback-only port. It upgraded the database in under a minute, kept the site, and the automation login worked. The rehearsal was then removed.
05:30–05:44Inventory pinned to 6.3.0.45, with 6.2 and 6.1 declared as fallbacks (inventory #24); role default set (builder #13); runbook written (docs #30)

The controller was down for about two minutes. No device was adopted, so nothing on site depended on it being up.

Found along the way: the controller had never adopted the switch. The AP had been pending since 2026-03-24, when it was forgotten with a configuration reset.

Follow-ups #

  • Decided who owns network device configuration: inventory owns it, and the controller applies it through its Open API — ADR-0009.
  • Fix or retire playbooks/upgrade-omada.yml in the builder collection. It hard-codes a fresh install and deletes the controller’s data.
Page last modified: September 10, 2026