[AI-3] Provision the LLM connection from the environment by tangopium · Pull Request #24887 · opf/openproject

tangopium · 2026-08-22T14:55:14Z

Ticket

What are you trying to accomplish?

PR 9 of 11 in the AI-3 stack. Adds the OPENPROJECT_LLM__CONNECTION setting and an EnvData seeder, so an on-premise deployment can ship the connection alongside its LLM sidecar without touching the UI. When the setting is present the record belongs to the deployment: the form renders read-only with a banner, the destructive actions refuse, and the contract rejects any write. Seeding must not fail because of a server that has not finished starting, so the environment path skips the reachability probe and refreshes the catalogue in a background job.

Merge checklist

Added/updated tests
Added/updated documentation (env var list regenerated)
Behind the llm_connection feature flag

Stacked on #24886.

github-actions · 2026-08-22T14:59:07Z

Caution

The provided work package version does not match the core version

Details:

Work package URL: https://community.openproject.org/work_packages/66020
Work package version: not set
Core version: 17.9.0

Please make sure that:

The work package version OR your pull request target branch is correct

github-actions · 2026-08-22T14:59:08Z

Caution

The Enterprise plan field is not set on the work package

Details:

Work package URL: https://community.openproject.org/work_packages/66020

Please make sure that:

The work package Enterprise plan field is set

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: dcefe085a3

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Adds the OPENPROJECT_LLM__CONNECTION setting and an EnvData seeder, so an on-premise deployment can ship the connection alongside its LLM sidecar without touching the UI. When the setting is present the record belongs to the deployment: the form renders read-only with a banner, the destructive actions refuse, and the base contract rejects any write. Seeding must not block on -- or fail because of -- a server that has not finished starting, so the environment path validates with its own contract that skips the reachability probe, and refreshes the catalogue through a background job instead of inline. Part 9 of the AI-3 stack. https://community.openproject.org/work_packages/66020

github-actions · 2026-08-22T16:14:04Z

Warning

Flaky specs

rspec ./spec/features/work_packages/table/queries/parent_field_ranking_spec.rb[1:1]
rspec ./spec/features/work_packages/table/switch_types_spec.rb[1:1:1]

🤖 Ask Copilot to investigate

Copy the prompt below into a new comment on this PR to delegate the investigation to GitHub Copilot. It will look into the flakiness and open a separate pull request with you as reviewer.

@copilot The following spec(s) are flaky in CI (first seen on PR #24887, linked for reference only):

- `rspec ./spec/features/work_packages/table/queries/parent_field_ranking_spec.rb[1:1]`
- `rspec ./spec/features/work_packages/table/switch_types_spec.rb[1:1:1]`

Treat this as a standalone task, unrelated to PR #24887. Create a new branch from origin/dev and open a new pull request targeting dev — do not stack it on PR #24887 or reuse that branch.

Follow the playbook in docs/development/testing/handling-flaky-tests/README.md to find the root cause and fix the underlying race — do not skip, delete, or weaken the spec to make it pass; disabling is a last resort per the playbook, and only with a bug ticket. Verify the fix by running the spec(s) repeatedly (e.g. `script/bulk_run_rspec --run-count 10`).

If you cannot reproduce the flake or are not confident in a fix after reasonable investigation, do not fabricate a change or skip the spec to force CI green. Instead, leave the pull request in draft and document what you tried, the suspected cause, and any leads in its description, then assign @tangopium to take over.

Once the fix is verified, title the PR after the spec(s) it fixes, and use the PR description to explain the root cause, how the change resolves it, and the before/after results. Label the PR `flaky-spec`, assign @tangopium, and request a review from @tangopium.
On every commit, set @tangopium as the sole co-author with a `Co-authored-by:` trailer (use their GitHub no-reply email so it links to their account), so it is traceable who dispatched the fix.

github-actions · 2026-08-23T01:06:33Z

Warning

Flaky specs

rspec ./modules/backlogs/spec/features/inbox_column_spec.rb[1:7:1]
rspec ./modules/meeting/spec/features/meeting_notifications_spec.rb[1:2:1]
rspec ./modules/meeting/spec/features/meeting_notifications_spec.rb[1:2:2]
rspec ./spec/features/roles/report_spec.rb[1:2]
rspec ./spec/features/work_packages/table/switch_types_spec.rb[1:1:2]

🤖 Ask Copilot to investigate

Copy the prompt below into a new comment on this PR to delegate the investigation to GitHub Copilot. It will look into the flakiness and open a separate pull request with you as reviewer.

@copilot The following spec(s) are flaky in CI (first seen on PR #24887, linked for reference only):

- `rspec ./modules/backlogs/spec/features/inbox_column_spec.rb[1:7:1]`
- `rspec ./modules/meeting/spec/features/meeting_notifications_spec.rb[1:2:1]`
- `rspec ./modules/meeting/spec/features/meeting_notifications_spec.rb[1:2:2]`
- `rspec ./spec/features/roles/report_spec.rb[1:2]`
- `rspec ./spec/features/work_packages/table/switch_types_spec.rb[1:1:2]`

Treat this as a standalone task, unrelated to PR #24887. Create a new branch from origin/dev and open a new pull request targeting dev — do not stack it on PR #24887 or reuse that branch.

Follow the playbook in docs/development/testing/handling-flaky-tests/README.md to find the root cause and fix the underlying race — do not skip, delete, or weaken the spec to make it pass; disabling is a last resort per the playbook, and only with a bug ticket. Verify the fix by running the spec(s) repeatedly (e.g. `script/bulk_run_rspec --run-count 10`).

If you cannot reproduce the flake or are not confident in a fix after reasonable investigation, do not fabricate a change or skip the spec to force CI green. Instead, leave the pull request in draft and document what you tried, the suspected cause, and any leads in its description, then assign @tangopium to take over.

Once the fix is verified, title the PR after the spec(s) it fixes, and use the PR description to explain the root cause, how the change resolves it, and the before/after results. Label the PR `flaky-spec`, assign @tangopium, and request a review from @tangopium.
On every commit, set @tangopium as the sole co-author with a `Co-authored-by:` trailer (use their GitHub no-reply email so it links to their account), so it is traceable who dispatched the fix.

The seeder runs on every container start, so enqueuing a refresh there undid whatever an administrator had curated, and could empty the list when the new server answers nothing. Following the UX review with Tom, the environment path now follows the same rule as the administration page: fetch the model list once while nothing is stored, and leave a stored list alone. A host or key changed in the environment shows up as the stale warning on the AI models page, which asks for the refresh.

The default chat model moved off the LLM settings page onto the AI models page in the UX review with Tom, and it moved out of reach of the environment banner with it. It now renders like the server settings do when the deployment owns the connection: the picker disabled, the banner above it, no Save button that the base contract would only reject. The autocompleter takes its disabled flag through autocomplete_options, which is what reaches the Angular component; the field-level flag it had before never left the Ruby side. The API key caption follows the same idea: asking an administrator to leave the field blank to keep the current key makes no sense where nothing can be typed, so it says where the key comes from instead.

The reviewer thread on the seeder was answered with the backoff retry, but nothing pinned it. The job now loops over every stored connection, so the example also proves that a server which is not up yet stops the run and comes back rather than silently leaving the connection without its model list.

The connection defaults reference a model row, so an environment naming a model the server has not been asked about yet had nothing to point at. The seed enters that model the way an administrator entering one by hand would, then writes the reference, both inside one transaction, and clears a default the environment no longer names. Provisioning also writes the AI features switch to its setting rather than to the column it used to live in, and the catalogue fill is now enqueued when no catalogue was ever fetched rather than when no model row exists, since provisioning itself creates one.

tangopium · 2026-09-10T07:22:58Z

The branch was rebuilt, so the commits your comments point at no longer exist and those threads now render as outdated. Nothing was dropped. Here is where each point landed.

The catalogue is filled once when provisioning, and the retry that a provisioned catalogue depends on is covered.
The default models are read-only while the environment owns the connection.
Because the defaults are references now, a model the environment names is entered as a row during provisioning, the way an administrator entering one by hand would. A default the environment stops naming is cleared.

CI has not started on the new commits: only the CLA check ran, and the workflow runs seem to need maintainer approval on this repository. The checks are waiting on that.

This was referenced Aug 22, 2026

[AI-3] Register AI features and resolve their models at runtime #24888

Open

[#66020] Add LLM connection settings to admin AI configuration #24851

Closed

chatgpt-codex-connector Bot reviewed Aug 22, 2026

View reviewed changes

Comment thread app/seeders/env_data/llm_connection_seeder.rb Outdated

tangopium force-pushed the feature/66020-pr09-environment-provisioning branch from dcefe08 to fa61b44 Compare August 22, 2026 15:05

tangopium changed the title ~~[#66020] Provision the LLM connection from the environment~~ [AI-3] Provision the LLM connection from the environment Aug 22, 2026

tangopium force-pushed the feature/66020-pr09-environment-provisioning branch from fa61b44 to 4ca02db Compare August 22, 2026 15:26

tangopium force-pushed the feature/66020-pr09-environment-provisioning branch from 4ca02db to b33ce35 Compare August 22, 2026 15:51

tangopium force-pushed the feature/66020-pr09-environment-provisioning branch from b33ce35 to 786a84f Compare August 22, 2026 16:16

tangopium requested a review from thykel August 22, 2026 21:39

tangopium self-assigned this Aug 22, 2026

tangopium added 4 commits September 7, 2026 15:41

tangopium force-pushed the feature/66020-pr09-environment-provisioning branch from 786a84f to 0c17575 Compare September 10, 2026 06:52

Navigation Menu

Uh oh!

[AI-3] Provision the LLM connection from the environment - #24887

[AI-3] Provision the LLM connection from the environment#24887
tangopium wants to merge 5 commits into
feature/66020-pr08-embedding-detectionfrom
feature/66020-pr09-environment-provisioning

tangopium commented Aug 22, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Aug 22, 2026

Uh oh!

github-actions Bot commented Aug 22, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

github-actions Bot commented Aug 22, 2026

Uh oh!

github-actions Bot commented Aug 23, 2026

Uh oh!

tangopium commented Sep 10, 2026

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

tangopium commented Aug 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Ticket

What are you trying to accomplish?

Merge checklist

Uh oh!

github-actions Bot commented Aug 22, 2026

Uh oh!

github-actions Bot commented Aug 22, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

github-actions Bot commented Aug 22, 2026

Uh oh!

github-actions Bot commented Aug 23, 2026

Uh oh!

tangopium commented Sep 10, 2026

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

1 participant

tangopium commented Aug 22, 2026 •

edited

Loading