feat(FR-2451): add auto scaling rule management feature spec by agatha197 · Pull Request #6356 · lablup/backend.ai-webui

agatha197 · 2026-04-02T15:19:12Z

Note

This is for v26.5.0

Resolves #6357 (FR-2451)

Summary

Add full-scope draft spec for Auto Scaling Rule management feature
Admin Serving page: Prometheus Query Preset CRUD for superadmins (create/read/update/delete via dedicated tab)
Service detail page: single/range condition mode selector, Prometheus preset-based metric selection, query result preview
Updated to reflect schema changes:
- Query APIs are user-accessible (prometheusQueryPresets, prometheusQueryPreset, prometheusQueryPresetResult), mutations remain admin-only
- AutoScalingRule.queryPreset: QueryDefinition field added (26.4.3) — enables direct preset resolution without ID roundtrip
- QueryDefinition.category: QueryPresetCategory field added (26.4.3) — preset grouping in UI dropdowns
- New category-related queries/mutations (prometheusQueryPresetCategories, adminCreatePrometheusQueryPresetCategory, etc.)
- QueryDefinition extended with description, rank, categoryId fields

Notes

This is the broader draft spec. The reduced, implementation-focused spec is in the parent PR feat(FR-2494): add auto scaling rule UX improvements spec #6486.
TODO(needs-backend): metric_source field may need version-based branching once the backend core auto scaling PR lands.

Spec

See `.specs/draft-auto-scaling-rule-management/spec.md`

agatha197 · 2026-04-02T15:19:25Z

feat(FR-2451): update spec to reflect schema changes (queryPreset field, QueryPresetCategory, category grouping) #6819
feat(FR-2451): add auto scaling rule management feature spec #6356 👈 (View in Graphite)
feat(FR-2494): add auto scaling rule UX improvements spec #6486 : 1 other dependent PR (#6639 )
main

How to use the Graphite Merge Queue

Add either label to this PR to merge it via the merge queue:

flow:merge-queue - adds this PR to the back of the merge queue
flow:hotfix - for urgent changes, fast-track this PR to the front of the merge queue

You must have a Graphite account in order to use the merge queue. Sign up using this link.

_{An organization admin has required the Graphite Merge Queue in this repository.} _{Please do not merge from GitHub as this will restart CI on PRs being processed by the merge queue.}

This stack of pull requests is managed by Graphite. Learn more about stacking.

Copilot

Pull request overview

Adds a draft feature specification for Auto Scaling Rule management in the Admin Serving area, covering Prometheus Query Preset CRUD (superadmin) and preset-based metric selection + condition mode enhancements in the service detail auto-scaling rule editor.

Changes:

Introduces a new spec document describing Admin Serving tab split and Preset CRUD flows.
Specifies service-detail modal UX changes (preset metric selection, single vs range condition mode).
Documents (current) dependency on an admin-only query result preview API.

…6459) Resolves #6453 ([FR-2483](https://lablup.atlassian.net/browse/FR-2483)) — sub-task of [FR-2470](https://lablup.atlassian.net/browse/FR-2470). - Fix the root-cause blocker where `EduAppLauncherPage` calls `useApiEndpoint()` but has never gone through login, so `apiEndpoint` resolves to an empty string - Resolve endpoint from `config.toml` at page level via `useEduAppApiEndpoint()` - Ensure the Backend.AI client is initialized before any downstream Relay query runs - `react/src/pages/EduAppLauncherPage.tsx` - `react/src/components/EduAppLauncher.tsx` - Technical Requirements #1 — `apiEndpoint` never empty - Must-Have: endpoint initialization with `wsproxy.proxyURL` - Out of Scope: `_token_login` untouched - [ ] Open EduAppLauncher page without prior login — endpoint should resolve from `config.toml` - [ ] No suspense error from Relay query firing before client is ready [FR-2483]: https://lablup.atlassian.net/browse/FR-2483?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ [FR-2470]: https://lablup.atlassian.net/browse/FR-2470?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ

…aged state machine (#6460) Resolves #6454 ([FR-2484](https://lablup.atlassian.net/browse/FR-2484)) — sub-task of [FR-2470](https://lablup.atlassian.net/browse/FR-2470). - Replace REST-based session lookup (`computeSession.list`) with a Relay `useLazyLoadQuery` over `ComputeSessionNode` so the returned fragment ref can feed `useBackendAIAppLauncher` in the next sub-task - Refactor `EduAppLauncher` into a staged state machine (auth → session → launch) - Add explicit session-creation error classification (resource shortage, missing template, 408 timeout, duplicate image session, other) — rendering lands in FR-2487 - `react/src/components/EduAppLauncher.tsx` - `react/src/__generated__/EduAppLauncher*.graphql.ts` (generated) - Technical Requirements #3 and #4 — session lookup via Relay, fragment data available - Acceptance Criteria: 5-bucket session-creation error classification (rendering is FR-2487) - Preserves existing token auth / reuse / creation flow - [ ] Existing session reuse path still works - [ ] New session creation path still works - [ ] State transitions traced via debug logs [FR-2484]: https://lablup.atlassian.net/browse/FR-2484?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ [FR-2470]: https://lablup.atlassian.net/browse/FR-2470?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ

…cation (#6462) Resolves #6456 ([FR-2486](https://lablup.atlassian.net/browse/FR-2486)) — sub-task of [FR-2470](https://lablup.atlassian.net/browse/FR-2470). - Remove the `CustomEvent('add-bai-notification')` workaround and the `NotificationForAnonymous` bridge in favor of the native `useSetBAINotification` hook - `EduAppLauncherPage` is already wrapped by antd `App` via `DefaultProvidersForReactRoot`, so `useSetBAINotification` works directly without a custom event hop - `react/src/components/EduAppLauncher.tsx` - `react/src/pages/EduAppLauncherPage.tsx` - Technical Requirements #5 and #6 — all notifications via `upsertNotification`; no `dispatchEvent('add-bai-notification')` in `EduAppLauncher.tsx` - Must-Have: remove `NotificationForAnonymous` dependency from `EduAppLauncherPage` - [ ] Success notifications show correctly - [ ] Error notifications show correctly (persistent for errors) - [ ] No stale CustomEvent listeners remain [FR-2486]: https://lablup.atlassian.net/browse/FR-2486?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ [FR-2470]: https://lablup.atlassian.net/browse/FR-2470?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ

Resolves #6457 ([FR-2487](https://lablup.atlassian.net/browse/FR-2487)) — sub-task of [FR-2470](https://lablup.atlassian.net/browse/FR-2470). - Render a centered Ant Design `Card` with `Steps` (인증 확인 → 세션 확인/생성 → 앱 실행) that visualizes progress, per-step errors, and the final completion state - Map classified session-creation errors from FR-2484 to per-step messages - Map service-port-missing errors from the hook to an 앱 실행 step error - Open the launched app in a new window (`window.open(..., '_blank')`), leaving the original page on the completion message - `react/src/components/EduAppLauncher.tsx` - `resources/i18n/*.json` (minimal additions) - UI/UX Acceptance Criteria (all 5) - Technical Acceptance: error classification rendering - Technical Acceptance: `service_ports` missing message - Must-Have: new window + completion state - [ ] Steps render correctly at each stage - [ ] Error at each step shows the correct classified message - [ ] On success, new window opens and page shows completion state [FR-2487]: https://lablup.atlassian.net/browse/FR-2487?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ [FR-2470]: https://lablup.atlassian.net/browse/FR-2470?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ

…ning (#6498) Resolves #6497 ([FR-2496](https://lablup.atlassian.net/browse/FR-2496)) — sub-task of [FR-2470](https://lablup.atlassian.net/browse/FR-2470). Follow-up polish on the EduAppLauncher refactor stack. Three related improvements surfaced during manual testing of the completed FR-2483 → FR-2487 work. When the launch completes, the original code only auto-opens the app window. Browsers with popup blockers silently drop the `window.open` call. This PR: - Stores the resolved `appConnectUrl` in the `done` stage so the UI knows where to navigate - Adds a visible primary button ("If a new window did not open automatically, click here.") on the completion card - Adds a secondary "Refresh page" action - Remove manual `useCallback` / `useMemoizedFn` wrappers now that the file uses the `'use memo'` directive (React Compiler memoizes automatically) - Wrap the launch-effect body in `useEffectEvent` so the `useEffect` deps stay precisely `[active, apiEndpoint]` — no more `// eslint-disable-next-line react-hooks/exhaustive-deps` - The `onLaunchEffect` block is placed **after** `_launch` is declared to respect the temporal-dead-zone rule (caught by `react-hooks/immutability` lint) Root-cause fix: the Backend.AI client initializes `_config.proxyURL = null` by default, and `config.toml` can ship `wsproxy.proxyURL = ""` for environments that rely on the local default. The previous `!== undefined` check let both null and empty-string pass through, which later crashed on `url.endsWith(...)`. Switch to a truthy check so both fall through to the default `http://127.0.0.1:5050/`. Also adds structured debug logging through the wsproxy launch path (`getWSProxyVersion`, `getProxyURL`, `_launchApp`) so the version/URL/wsproxy response chain is observable from the browser console. New keys `eduAppLauncher.OpenAppInNewWindow` and `eduAppLauncher.RefreshPage` added in all 21 supported languages. Codifies when to reach for `useEffectEvent` over `useCallback` under the `'use memo'` directive. Captures the temporal-dead-zone gotcha that this PR itself tripped on. SSH/VS Code Desktop modal apps, Edu API runtime state handling (inherited from FR-2470 out-of-scope list). - [ ] With a popup-blocker enabled: completion card shows the fallback button and it navigates correctly on click - [ ] Without a popup-blocker: original `window.open` path still works; completion card still renders the fallback as a recovery option - [ ] `config.toml` with `wsproxy.proxyURL = ""` no longer crashes; falls through to `http://127.0.0.1:5050/` - [ ] `config.toml` with a real proxy URL still honors it - [ ] Browser console shows the new `[wsproxy]` / `[_launchApp]` logs at `info` level - [ ] No `useCallback` / `useMemoizedFn` regressions in `EduAppLauncher.tsx` [FR-2496]: https://lablup.atlassian.net/browse/FR-2496?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ [FR-2470]: https://lablup.atlassian.net/browse/FR-2470?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ

…6452) Resolves #6443 (FR-2467) - Disable commit button when session status is not RUNNING - Previously only checked isActive() which allowed non-running states like PREPARING, PENDING, PULLING to enable the button - [ ] Verify commit button is disabled when session is in PREPARING state - [ ] Verify commit button is disabled when session is in PENDING state - [ ] Verify commit button is enabled when session is in RUNNING state - [ ] Verify commit button remains disabled for non-owners regardless of session status 🤖 Generated with [Claude Code](https://claude.com/claude-code)

…ld, QueryPresetCategory, category grouping)

github-actions Bot assigned agatha197 Apr 2, 2026

github-actions Bot added the size:L 100~500 LoC label Apr 2, 2026

agatha197 changed the title ~~feat: add auto scaling rule management feature spec~~ feat(FR-2451): add auto scaling rule management feature spec Apr 2, 2026

agatha197 marked this pull request as ready for review April 2, 2026 15:25

Copilot AI review requested due to automatic review settings April 2, 2026 15:25

agatha197 marked this pull request as draft April 2, 2026 15:25

Copilot started reviewing on behalf of agatha197 April 2, 2026 15:25 View session

agatha197 requested a review from yomybaby April 2, 2026 15:26

Copilot AI reviewed Apr 2, 2026

View reviewed changes

agatha197 force-pushed the 04-03-feat_add_auto_scaling_rule_management_feature_spec branch 3 times, most recently from ad4647a to 1314f5f Compare April 7, 2026 08:12

agatha197 changed the base branch from main to graphite-base/6356 April 7, 2026 08:53

agatha197 force-pushed the 04-03-feat_add_auto_scaling_rule_management_feature_spec branch from 1314f5f to 671fe73 Compare April 7, 2026 08:53

agatha197 changed the base branch from graphite-base/6356 to 04-07-feat_fr-2494_add_auto_scaling_rule_ux_improvements_spec April 7, 2026 08:53

agatha197 mentioned this pull request Apr 7, 2026

feat(FR-2494): add auto scaling rule UX improvements spec #6486

Merged

agatha197 changed the base branch from 04-07-feat_fr-2494_add_auto_scaling_rule_ux_improvements_spec to graphite-base/6356 April 7, 2026 09:11

agatha197 force-pushed the graphite-base/6356 branch from a6b71eb to c20cb58 Compare April 7, 2026 09:31

agatha197 force-pushed the 04-03-feat_add_auto_scaling_rule_management_feature_spec branch 2 times, most recently from 52937eb to d147436 Compare April 9, 2026 02:34

agatha197 force-pushed the graphite-base/6356 branch from c20cb58 to 60b5a7a Compare April 9, 2026 02:34

agatha197 changed the base branch from graphite-base/6356 to main April 21, 2026 04:58

agatha197 mentioned this pull request Apr 21, 2026

feat(FR-2451): update spec to reflect schema changes (queryPreset field, QueryPresetCategory, category grouping) #6819

Closed

5 tasks

github-actions Bot added size:XL 500~ LoC and removed size:L 100~500 LoC labels Apr 21, 2026

agatha197 marked this pull request as ready for review April 21, 2026 05:07

agatha197 force-pushed the 04-03-feat_add_auto_scaling_rule_management_feature_spec branch from 3232f2c to 7f6e62e Compare April 21, 2026 06:26

github-actions Bot added size:L 100~500 LoC and removed size:XL 500~ LoC labels Apr 21, 2026

agatha197 and others added 8 commits April 23, 2026 10:31

feat(FR-2494): add auto scaling rule UX improvements spec

88115fe

feat: add auto scaling rule management feature spec

ff61903

agatha197 force-pushed the 04-03-feat_add_auto_scaling_rule_management_feature_spec branch from 7f6e62e to f2f0030 Compare April 23, 2026 01:42

agatha197 marked this pull request as draft April 23, 2026 01:44

feat(FR-2451): update spec to reflect schema changes (queryPreset fie…

5dea801

…ld, QueryPresetCategory, category grouping)

agatha197 force-pushed the 04-03-feat_add_auto_scaling_rule_management_feature_spec branch from f2f0030 to 5dea801 Compare April 23, 2026 01:58

agatha197 marked this pull request as ready for review April 23, 2026 01:59

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(FR-2451): add auto scaling rule management feature spec#6356

feat(FR-2451): add auto scaling rule management feature spec#6356
agatha197 wants to merge 9 commits intomainfrom
04-03-feat_add_auto_scaling_rule_management_feature_spec

agatha197 commented Apr 2, 2026 •

edited

Loading

Uh oh!

agatha197 commented Apr 2, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

agatha197 commented Apr 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Notes

Spec

Uh oh!

agatha197 commented Apr 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

How to use the Graphite Merge Queue

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

agatha197 commented Apr 2, 2026 •

edited

Loading

agatha197 commented Apr 2, 2026 •

edited

Loading