feat(settings): add endpoint and UI for switching media server#2539
Draft
feat(settings): add endpoint and UI for switching media server#2539
Conversation
📝 WalkthroughWalkthroughAdds a media-server switching feature: new POST API to switch servers, server-side migration logic updating Settings, Users, Media, Watchlists, and Sessions, updated Plex/Jellyfin/Emby auth/linking flows, new admin UI to initiate switches, and a stricter Media update guard. Changes
Sequence Diagram(s)sequenceDiagram
participant User as User
participant UI as Settings UI
participant API as Settings API
participant DB as Database
participant BG as Background Jobs
User->>UI: Select target server & confirm
UI->>API: POST /settings/switch-media-server { targetServerType }
API->>API: Validate current settings & tokens
API->>DB: Begin transaction / update Settings
rect rgba(100,150,200,0.5)
API->>DB: Update User records (clear/set server fields)
API->>DB: Reset Media records (clear server-specific IDs)
API->>DB: Reset Watchlist entries
API->>DB: Purge Sessions
end
API->>DB: Commit transaction
API->>BG: Restart scheduled background jobs
BG->>BG: Reinitialize with new server config
API-->>UI: 200 Success
UI->>User: Show success toast and reload
Estimated code review effort🎯 4 (Complex) | ⏱️ ~60 minutes Possibly related PRs
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 4✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Contributor
Author
|
@coderabbitai review |
✅ Actions performedReview triggered.
|
0xSysR3ll
commented
Feb 20, 2026
Contributor
Author
There was a problem hiding this comment.
Don't mind the big refactor on this one, I had some UI issues that generated a lot of changes (it's mainly spaces)
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 6
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
server/routes/settings/index.ts (1)
175-219:⚠️ Potential issue | 🟡 MinorHandle missing Plex token explicitly.
If the admin token is absent (e.g., post-switch), this returns
undefinedwith 200. A clear 400 response is safer.💡 Suggested fix
- const authToken = admin.plexToken ?? null; - const plexTvClient = authToken ? new PlexTvAPI(authToken) : null; + const authToken = admin.plexToken ?? null; + if (!authToken) { + return next({ + status: 400, + message: 'Plex token missing. Sign in with Plex first.', + }); + } + const plexTvClient = new PlexTvAPI(authToken);🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@server/routes/settings/index.ts` around lines 175 - 219, The route currently proceeds when the admin Plex token is missing and later returns undefined; before creating PlexTvAPI/calling plexTvClient.getDevices(), check admin.plexToken (authToken) and if null/undefined immediately return a 400 Bad Request response (with a clear message like "Plex token missing") instead of continuing; update the logic around getSettings / userRepository.findOneOrFail / PlexTvAPI / plexTvClient to short-circuit on missing authToken so no calls to plexTvClient.getDevices() or PlexAPI are attempted.
🧹 Nitpick comments (1)
seerr-api.yml (1)
2293-2320: Document the optional target server type in the request body.Clients won’t know how to request Emby vs. Jellyfin without a schema. Consider adding an optional
targetServerTypefield.📝 Suggested OpenAPI addition
/settings/switch-media-server: post: summary: Switch media server tags: - settings + requestBody: + required: false + content: + application/json: + schema: + type: object + properties: + targetServerType: + type: string + enum: [jellyfin, emby] + description: Target server type; defaults to jellyfin when omitted. responses: '200': description: Media server cleared🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@seerr-api.yml` around lines 2293 - 2320, The POST /settings/switch-media-server operation lacks a requestBody schema describing how to specify which server type to switch to; add a requestBody (content: application/json) with an optional property named targetServerType (string) and constrain it to allowed values (e.g., enum: ["emby","jellyfin"]) so clients know how to request Emby vs Jellyfin; update the operation under the path "/settings/switch-media-server" to include this requestBody and mark the property as optional and documented in the response examples.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@server/routes/auth.ts`:
- Around line 61-92: The current block lets any authenticated user overwrite the
admin Plex token; update the authorization to allow only admins to perform this
action by checking the authenticated user's admin role (e.g., req.user.isAdmin
or req.user.role === 'admin') before proceeding with PlexTvAPI validation and
saving to userRepository; if the user is not an admin, return a 403 error (do
not change the existing admin user retrieval logic that uses findOneOrFail for
id 1 and the PlexTvAPI validation flow).
In `@server/routes/settings/index.ts`:
- Around line 453-569: The /switch-media-server handler
(settingsRoutes.post('/switch-media-server', ...)) is unprotected; add an
admin-only guard at the top of the handler (or attach an existing admin
middleware) so only admins can call it: verify req.user exists and has admin
rights (e.g., check req.user.userType === UserType.ADMIN or an isAdmin flag on
the User entity) and return 403 if not authorized before any state changes
(before getSettings(), repository updates, settings.save(), startJobs(), etc.);
reference User and UserType used in the route for the check or reuse your app's
ensureAdmin middleware if available.
In `@server/routes/user/usersettings.ts`:
- Around line 419-429: The unlink (Jellyfin delete) route currently blocks
unlinking when Jellyfin/Emby isn't the main server; update its guard to mirror
the linking check by using the same logic as isMainJellyfin and
jellyfinForLinking: allow unlink if either settings.main.mediaServerType ===
MediaServerType.JELLYFIN/EMBY (isMainJellyfin) or settings.jellyfin?.ip is
present; replace the strict isMainJellyfin-only check in the delete handler with
the combined condition (!isMainJellyfin && !jellyfinForLinking?.ip) so users who
linked against a configured non-main Jellyfin can unlink.
In `@src/components/Settings/SettingsPlex.tsx`:
- Around line 792-801: The Formik form for Tautulli (the Formik component with
initialValues built from dataTautulli and using TautulliSettingsSchema) can
mount before dataTautulli is available and won't pick up updates; enable
reinitialization by adding enableReinitialize to the Formik props (or
conditionally render Formik only when dataTautulli is defined) so the form's
initialValues update when dataTautulli arrives and avoid blank
defaults/accidental overwrites.
In `@src/components/Settings/SwitchMediaServerSection.tsx`:
- Around line 107-113: In the catch block inside SwitchMediaServerSection (the
try/catch that builds the toast message), broaden the Axios error extraction to
check response.data.message as well as response.data.error and fallback to
err.message before using the generic intl.formatMessage; update the conditional
that sets message (currently using axios.isAxiosError(err) &&
err.response?.data?.error) to instead test axios.isAxiosError(err) and then
prefer String(err.response?.data?.error ?? err.response?.data?.message ??
err.message) so you capture endpoints that return message or a plain Error
message while preserving the existing generic fallback.
- Around line 95-106: The success toast message is hard-coded; replace it with a
localized string by calling the project's i18n translator instead of the literal
text. In the onClick handler inside SwitchMediaServerSection (where addToast is
called after axios.post), import or use the existing translation function (e.g.,
useTranslation's t or i18n.t) and pass a translation key like
t('settings.mediaServer.cleared') (with an appropriate entry added to your
locale files) to addToast instead of the hard-coded "Media server cleared. You
may need to restart."; keep the same appearance options and reload logic.
---
Outside diff comments:
In `@server/routes/settings/index.ts`:
- Around line 175-219: The route currently proceeds when the admin Plex token is
missing and later returns undefined; before creating PlexTvAPI/calling
plexTvClient.getDevices(), check admin.plexToken (authToken) and if
null/undefined immediately return a 400 Bad Request response (with a clear
message like "Plex token missing") instead of continuing; update the logic
around getSettings / userRepository.findOneOrFail / PlexTvAPI / plexTvClient to
short-circuit on missing authToken so no calls to plexTvClient.getDevices() or
PlexAPI are attempted.
---
Nitpick comments:
In `@seerr-api.yml`:
- Around line 2293-2320: The POST /settings/switch-media-server operation lacks
a requestBody schema describing how to specify which server type to switch to;
add a requestBody (content: application/json) with an optional property named
targetServerType (string) and constrain it to allowed values (e.g., enum:
["emby","jellyfin"]) so clients know how to request Emby vs Jellyfin; update the
operation under the path "/settings/switch-media-server" to include this
requestBody and mark the property as optional and documented in the response
examples.
0xSysR3ll
force-pushed
the
0xsysr3ll/feat/media-server-migration
branch
from
41a7f8b to
3786a27
Compare
github-actions
bot
added
the
i18n-out-of-sync
|
The i18n check failed because translation messages are out of sync. This usually happens when you've added or modified translation strings in your code but haven't updated the translation file. Please run |
github-actions
bot
removed
the
i18n-out-of-sync
Contributor
Author
|
@coderabbitai review |
✅ Actions performedReview triggered.
|
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (2)
server/subscriber/MediaSubscriber.ts (1)
155-158: Guard logic is correct; consider removing redundant optional chaining.The guard now ensures both
event.entityandevent.databaseEntityare defined. However, lines 178 and 191 still useevent.databaseEntity?.statusandevent.databaseEntity?.status4k— the optional chaining is now redundant and could be simplified for consistency.♻️ Optional cleanup
if ( - (event.entity.status !== event.databaseEntity?.status || + (event.entity.status !== event.databaseEntity.status || (event.entity.mediaType === MediaType.TV && seasonStatusCheck(false))) && validStatuses.includes(event.entity.status)if ( - (event.entity.status4k !== event.databaseEntity?.status4k || + (event.entity.status4k !== event.databaseEntity.status4k || (event.entity.mediaType === MediaType.TV && seasonStatusCheck(true))) && validStatuses.includes(event.entity.status4k)🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@server/subscriber/MediaSubscriber.ts` around lines 155 - 158, The guard in MediaSubscriber.afterUpdate already ensures event.entity and event.databaseEntity are non-null, so remove the redundant optional chaining: replace occurrences of event.databaseEntity?.status and event.databaseEntity?.status4k with direct access event.databaseEntity.status and event.databaseEntity.status4k (e.g., in the conditional checks currently using ?. at lines referencing those properties) while keeping the existing early-return guard intact.server/routes/settings/index.ts (1)
479-492: Consider: User update queries always match all users.The WHERE clause
user.id >= :zerowithzero: 0will always match all users since IDs are positive integers. While this works correctly, it's semantically unclear.Consider simplifying to make the intent explicit:
♻️ Suggested clarification
await userRepository .createQueryBuilder() .update(User) .set({ plexId: null, plexUsername: null, plexToken: null }) - .where('user.id >= :zero', { zero: 0 }) + .execute(); // Updates all usersOr if you need a WHERE clause for safety:
- .where('user.id >= :zero', { zero: 0 }) + .where('user.id IS NOT NULL') // All users🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@server/routes/settings/index.ts` around lines 479 - 492, The first update query uses a redundant WHERE ('user.id >= :zero', { zero: 0 }) that effectively matches all users; remove that semantic noise by calling the same userRepository.createQueryBuilder().update(User).set({ plexId: null, plexUsername: null, plexToken: null }).execute() so the intent to update all rows is explicit and clear (keep the second query using .where('user.jellyfinUserId IS NOT NULL') as-is).
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@server/routes/settings/index.ts`:
- Around line 500-506: The watchlist clearing is asymmetric and uses
inconsistent null handling: update the Watchlist clearing logic that uses
watchlistRepository.createQueryBuilder().update(Watchlist) so that
Watchlist.ratingKey is cleared whenever switching away from Plex (add the
corresponding branch or call in the switch-to-Plex/from-Plex flows), and make
the clearing consistent with other media fields by setting ratingKey to null
instead of ''. Replace .set({ ratingKey: '' }) with .set({ ratingKey: null })
and change the predicate to .where("watchlist.ratingKey IS NOT NULL") (or
equivalent ORM null check), and ensure the same null convention is applied
wherever media fields are cleared.
---
Nitpick comments:
In `@server/routes/settings/index.ts`:
- Around line 479-492: The first update query uses a redundant WHERE ('user.id
>= :zero', { zero: 0 }) that effectively matches all users; remove that semantic
noise by calling the same userRepository.createQueryBuilder().update(User).set({
plexId: null, plexUsername: null, plexToken: null }).execute() so the intent to
update all rows is explicit and clear (keep the second query using
.where('user.jellyfinUserId IS NOT NULL') as-is).
In `@server/subscriber/MediaSubscriber.ts`:
- Around line 155-158: The guard in MediaSubscriber.afterUpdate already ensures
event.entity and event.databaseEntity are non-null, so remove the redundant
optional chaining: replace occurrences of event.databaseEntity?.status and
event.databaseEntity?.status4k with direct access event.databaseEntity.status
and event.databaseEntity.status4k (e.g., in the conditional checks currently
using ?. at lines referencing those properties) while keeping the existing
early-return guard intact.
2 tasks
0xSysR3ll
force-pushed
the
0xsysr3ll/feat/media-server-migration
branch
2 times, most recently
from
99ba4fb to
ce959d6
Compare
… with existing users
… Emby as main server
0xSysR3ll
force-pushed
the
0xsysr3ll/feat/media-server-migration
branch
from
ce959d6 to
bec2302
Compare
Contributor
Author
|
@coderabbitai review |
✅ Actions performedReview triggered.
|
There was a problem hiding this comment.
Actionable comments posted: 6
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
server/routes/settings/index.ts (1)
126-160:⚠️ Potential issue | 🟠 MajorValidate against a temporary Plex config first.
Object.assign(settings.plex, plexBody)mutates live process state beforegetStatus()succeeds. If the connection test fails, the request returns an error but the app keeps serving the unsaved, unverified Plex host/port until something overwrites it. Mirror the Jellyfin path here and only copy intosettings.plexafter validation passes.Proposed fix
const settings = getSettings(); const body = req.body as Record<string, unknown>; const { authToken: bodyToken, ...plexBody } = body; try { const admin = await userRepository.findOneOrFail({ select: { id: true, plexToken: true }, where: { id: 1 }, }); - Object.assign(settings.plex, plexBody); + const tempPlexSettings = { ...settings.plex, ...plexBody }; const token = settings.main.mediaServerType !== MediaServerType.PLEX && typeof bodyToken === 'string' ? bodyToken : admin.plexToken; @@ - const plexClient = new PlexAPI({ plexToken: token }); + const plexClient = new PlexAPI({ + plexToken: token, + plexSettings: tempPlexSettings, + }); const result = await plexClient.getStatus(); @@ - settings.plex.machineId = result.MediaContainer.machineIdentifier; - settings.plex.name = result.MediaContainer.friendlyName; + Object.assign(settings.plex, tempPlexSettings, { + machineId: result.MediaContainer.machineIdentifier, + name: result.MediaContainer.friendlyName, + }); await settings.save();🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@server/routes/settings/index.ts` around lines 126 - 160, The current code mutates live settings via Object.assign(settings.plex, plexBody) before validating the Plex connection; instead, create a temporary plex config (e.g., const tempPlex = { ...settings.plex, ...plexBody }) and use that when constructing PlexAPI and calling plexClient.getStatus(); only after getStatus() confirms MediaContainer.machineIdentifier should you assign the validated values into settings.plex and call settings.save(); ensure the fallback token logic still uses bodyToken/admin.plexToken when building PlexAPI so the validation uses the intended credentials.src/components/UserProfile/UserSettings/UserLinkedAccountsSettings/index.tsx (1)
78-90:⚠️ Potential issue | 🟠 MajorDon't hide non-main Jellyfin links behind
userType.When Jellyfin/Emby is linked while another server is still primary, the server-side link flow can leave
user.userTypeunchanged. A local user that links here will therefore drop out ofaccounts, and the dropdown will still offer the link action instead of showing the linked account/unlink control. Please render secondary Jellyfin links from the stored linked fields, not just fromuserType.Also applies to: 136-141
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/components/UserProfile/UserSettings/UserLinkedAccountsSettings/index.tsx` around lines 78 - 90, The code currently gates adding Emby/Jellyfin to accounts on user.userType, causing linked secondary servers to be hidden; update the logic in the component that builds the accounts array so it relies on the stored linked fields (e.g., user.jellyfinUsername and user.embyUsername) rather than user.userType: always push a LinkedAccountType.Jellyfin entry when user.jellyfinUsername exists, and push a LinkedAccountType.Emby entry when user.embyUsername (or the correct Emby username field) exists, removing the user.userType checks around these pushes and ensuring you don't push duplicate entries for the same username.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@server/routes/auth.ts`:
- Around line 73-89: The handler currently overwrites the canonical Plex account
by always writing the validated Plex identity to the row found with
userRepository.findOneOrFail({ where: { id: 1 } }) (variables PlexTvAPI,
plextv.getUser, admin), allowing any admin to replace the system-wide Plex
token; change the guard so only the canonical owner can update that row: verify
req.user.id === 1 (or a dedicated owner flag on req.user) before assigning
admin.plexToken/plexId/plexUsername and saving via userRepository.save(admin);
alternatively, persist the validated token to the calling user's record instead
of id=1, or add a stable "isPrimary" flag on users and restrict writes to the
user where isPrimary === true and req.user is that same user. Ensure
req.user.hasPermission(Permission.ADMIN) remains but add the identity check
(req.user.id === 1 or req.user.isPrimary) before mutating the canonical account.
In `@server/routes/settings/index.ts`:
- Around line 561-572: The code calls setImmediate(() => startJobs()) after
clearing sessions which leads to duplicate schedulers because startJobs()
appends into scheduledJobs without clearing or cancelling existing jobs; update
the logic in the settings route (the branch that calls startJobs()) to first
cancel/clear existing scheduled jobs (e.g., iterate scheduledJobs and call their
cancel/clear method then empty the scheduledJobs array) or remove the
setImmediate call and defer starting jobs until process boot; apply the same fix
to the other occurrence around the startJobs() call at the 641-650 region to
ensure no duplicate schedulers are created.
- Around line 531-565: Wrap all DB mutations and the settings persistence into a
single database transaction so the media-server switch is atomic: perform the
repository updates currently using getRepository(User), getRepository(Media),
getRepository(Watchlist), the Session delete, and the settings.save() call
inside a single transactional callback (e.g., getManager().transaction or
getConnection().transaction) and only update the in-memory
settings.main.mediaServerType and settings.plex (and any other in-memory state)
after the transaction successfully commits; if the transaction fails, roll back
and do not modify in-memory values. Ensure the same change is applied for the
other branch mentioned (lines 603-645) so both switch paths use the transaction.
In `@server/routes/user/usersettings.ts`:
- Around line 339-345: The unlink handlers currently set user.userType =
UserType.LOCAL unconditionally, which downgrades accounts even when removing a
non-active (auxiliary) provider; change the logic to only set user.userType =
UserType.LOCAL when the provider being unlinked is the active auth provider
(compare settings.main.mediaServerType / MediaServerType.PLEX and the provider
being removed), otherwise leave user.userType unchanged and only remove the
provider-specific fields (e.g., plex fields checked by plexConfigured and
similar Jellyfin/Emby checks); update both unlink code paths (the Plex handler
around plexConfigured and the other handler referenced at lines ~508-516) to
perform this conditional fallback.
In `@src/components/Settings/SettingsPlex.tsx`:
- Around line 425-468: The onSubmit handler currently calls syncLibraries()
fire-and-forget which can lead to unhandled rejections and incorrect toasts;
modify the onSubmit block to await syncLibraries() so the save success toast
only appears after sync completion, or change the syncLibraries implementation
to ensure it always resets state by wrapping its body in try/finally
(setIsSyncing(true) before, finally setIsSyncing(false)) and await internal
axios.get('/api/v1/settings/plex/library') and revalidate() so errors are
handled and isSyncing is cleared.
In `@src/components/UserList/index.tsx`:
- Around line 731-745: Add plexId?: number | null and jellyfinUserId?: string |
null to the shared User interface in useUser (and adjust User.filteredFields /
User.filterMany() behavior if these fields are intentionally admin-only), then
remove the unsafe ad-hoc casts in the UserList component that check (user as {
plexId?: number | null }).plexId and (user as { jellyfinUserId?: string | null
}).jellyfinUserId; if these fields are only present for users with MANAGE_USERS
permission, document that constraint in a comment near the User interface and
ensure filteredFields/filterMany() consistently strips them for non-admins so
the badges render only when the field is actually present.
---
Outside diff comments:
In `@server/routes/settings/index.ts`:
- Around line 126-160: The current code mutates live settings via
Object.assign(settings.plex, plexBody) before validating the Plex connection;
instead, create a temporary plex config (e.g., const tempPlex = {
...settings.plex, ...plexBody }) and use that when constructing PlexAPI and
calling plexClient.getStatus(); only after getStatus() confirms
MediaContainer.machineIdentifier should you assign the validated values into
settings.plex and call settings.save(); ensure the fallback token logic still
uses bodyToken/admin.plexToken when building PlexAPI so the validation uses the
intended credentials.
In
`@src/components/UserProfile/UserSettings/UserLinkedAccountsSettings/index.tsx`:
- Around line 78-90: The code currently gates adding Emby/Jellyfin to accounts
on user.userType, causing linked secondary servers to be hidden; update the
logic in the component that builds the accounts array so it relies on the stored
linked fields (e.g., user.jellyfinUsername and user.embyUsername) rather than
user.userType: always push a LinkedAccountType.Jellyfin entry when
user.jellyfinUsername exists, and push a LinkedAccountType.Emby entry when
user.embyUsername (or the correct Emby username field) exists, removing the
user.userType checks around these pushes and ensuring you don't push duplicate
entries for the same username.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 2f42b3ff-f404-4821-9ebb-ac6e2bb3d247
📒 Files selected for processing (13)
seerr-api.ymlserver/middleware/deprecation.tsserver/routes/auth.tsserver/routes/settings/index.tsserver/routes/user/usersettings.tsserver/subscriber/MediaSubscriber.tssrc/components/Settings/SettingsLayout.tsxsrc/components/Settings/SettingsMain/index.tsxsrc/components/Settings/SettingsPlex.tsxsrc/components/Settings/SwitchMediaServerSection.tsxsrc/components/UserList/index.tsxsrc/components/UserProfile/UserSettings/UserLinkedAccountsSettings/index.tsxsrc/i18n/locale/en.json
✅ Files skipped from review due to trivial changes (1)
- server/middleware/deprecation.ts
🚧 Files skipped from review as they are similar to previous changes (2)
- src/components/Settings/SettingsLayout.tsx
- seerr-api.yml
Comment on lines
+73
to
+89
| if (!req.user.hasPermission(Permission.ADMIN)) { | ||
| return next({ | ||
| status: 403, | ||
| message: 'Admin permissions required.', | ||
| }); | ||
| } | ||
| try { | ||
| const plextv = new PlexTvAPI(body.authToken); | ||
| const account = await plextv.getUser(); | ||
| const admin = await userRepository.findOneOrFail({ | ||
| where: { id: 1 }, | ||
| }); | ||
| admin.plexToken = body.authToken; | ||
| admin.plexId = account.id; | ||
| admin.plexUsername = account.username; | ||
| await userRepository.save(admin); | ||
| return res.status(200).json({ email: admin.email }); |
There was a problem hiding this comment.
Don't let secondary admins overwrite the canonical Plex account.
Line 82 always writes the validated Plex identity onto user id = 1, and Lines 140-145 later reuse that row as mainUser for Plex access checks. That means any non-owner admin can replace the system-wide Plex account with their own token and change who Seerr treats as the canonical Plex user.
🔒 Minimal guard if `id = 1` remains the canonical storage
if (!req.user.hasPermission(Permission.ADMIN)) {
return next({
status: 403,
message: 'Admin permissions required.',
});
}
+ if (req.user.id !== 1) {
+ return next({
+ status: 403,
+ message: 'Owner permissions required.',
+ });
+ }
try {
const plextv = new PlexTvAPI(body.authToken);🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@server/routes/auth.ts` around lines 73 - 89, The handler currently overwrites
the canonical Plex account by always writing the validated Plex identity to the
row found with userRepository.findOneOrFail({ where: { id: 1 } }) (variables
PlexTvAPI, plextv.getUser, admin), allowing any admin to replace the system-wide
Plex token; change the guard so only the canonical owner can update that row:
verify req.user.id === 1 (or a dedicated owner flag on req.user) before
assigning admin.plexToken/plexId/plexUsername and saving via
userRepository.save(admin); alternatively, persist the validated token to the
calling user's record instead of id=1, or add a stable "isPrimary" flag on users
and restrict writes to the user where isPrimary === true and req.user is that
same user. Ensure req.user.hasPermission(Permission.ADMIN) remains but add the
identity check (req.user.id === 1 or req.user.isPrimary) before mutating the
canonical account.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Caution
Work in progress
This feature is still experimental. Do not use it on a production instance unless you have a verified backup and a tested rollback plan.
See Backups.
Also, don't use this image if you did not migrate to Seerr 3.1.0
This PR adds an in‑app way to switch a single Seerr instance between media servers (Plex <=> Jellyfin/Emby) without losing users, permissions, requests, or general settings.
The same database and application instance are reused; only the active media server and authentication method change.
The goal is to make media‑server migrations practical without needing a second Seerr instance or manual database edits. You (the owner) configure the new server, users link their new accounts, and then the instance is switched in place.
Note
This will initially ship as a preview so the flow can be tested and refined properly.
Migration flow (preview)
Back up first
Switch to the preview image
preview-media-server-migrationConfigure your target media server
Have users link their accounts
Switch media server in settings
Restart and sign in again
Verify your data
Go back to production image
latestor3.1.0How Has This Been Tested?
=> And vice-versa from Jellyfin to Plex.
Screenshots / Logs (if applicable)
Checklist:
pnpm buildpnpm i18n:extractSummary by CodeRabbit
New Features
Improvements
Localization