mirror of
https://github.com/Studio-Saelix/sencho.git
synced 2026-07-28 04:38:59 +00:00
Anso
0be479100c
* docs(introduction): rewrite intro page and refresh screenshots
Rewrite the Getting Started introduction to reflect the current product:
adds the Mesh, Blueprints, Pilot, Fleet, Resources concepts up front;
restructures capability sections around Stacks, Fleet (now including
Fleet Actions), Mesh, Blueprints, Monitoring, Resources, Security,
Automation, and Pilot/Remote ops; cross-links every claim to the
matching feature page.
Replaces the dashboard hero shot with a fresh capture against the
redesigned cockpit chrome and adds three inline shots (running stack
with anatomy and logs, fleet command deck, resources hub with
reclaim header). All screenshots taken at 1920x900, dark theme,
with node names, usernames, IPs, and home paths neutralized.
Drops outdated claims: stale "190+ templates" count, the
"viewer accounts" RBAC summary, and the "atomic" deployment label
that did not match the actual rollback mechanism.
* docs(introduction): add Federation and Fleet Secrets
Federation has shipped (Admiral) with cordon and pin policy as
operator overrides on the blueprint reconciler. Fleet Secrets is
landing as a Skipper+ tab for versioned env-var bundles encrypted
at rest, with diff preview and target push.
Mirror those in the Run-one-machine-or-many bullet list and in the
Tiers paragraph so the introduction matches the current product.
* docs(introduction): re-shoot screenshots against v0.72.0 production
Re-capture all four introduction screenshots from the upgraded
production node so the fleet view shows the current full tab strip
(Overview / Snapshots / Status / Deployments / Traffic / Federation
/ Fleet Actions / Secrets) instead of the older Overview / Snapshots
/ Status only. Same 1920x900 dark-theme capture and the same PII
scrub applied (node names, usernames, IPs, home paths neutralized).
* docs(quickstart): rewrite around v0.72.0 cockpit and add screenshots
Replaces the bare install snippet with a five-minute walkthrough that
matches the redesigned UI. Leads with a docker-compose.yml block (the
bare docker run command is collapsed in an Accordion), keeps the 1:1
path rule, and adds two new sections that show the user what happens
on first boot.
Adds two screenshots at 1920x900 dark theme:
- setup-cold-start.png: the Cold start card with Username, Password,
Confirm password fields, and the Initialize console button.
- dashboard.png: the post-sign-in dashboard captured against v0.72.0
with the full top nav (Home, Fleet, Resources, App Store, Logs,
Auto-Update, Console, Audit, Schedules) and the populated Stack
health table sorted by load.
Where-to-next now uses CardGroup cols=2 to match the introduction
page's pattern.
* docs(configuration): align env var reference with current backend
Bring docs/getting-started/configuration.mdx up to date with the
v0.72.0 backend:
- Remove PORT from the optional env vars table. The listen port is
hardcoded to 1852 in backend/src/helpers/constants.ts and is never
read from the environment. Replace it with a Listen port section
that explains the fixed port and host-port remapping.
- Document API_RATE_LIMIT (default 200) and API_POLLING_RATE_LIMIT
(default 300), both applied in production only.
- Note the /app/compose fallback default for COMPOSE_DIR while still
pointing readers at the 1:1 path rule.
- Point the SSO env var section at the new SSO Quickstart page and
keep the SSO feature reference as the deeper dive.
- Tighten First boot and cross-link to Quickstart so the screenshot
is not duplicated.
- Add a Where to next CardGroup matching the refreshed Introduction
and Quickstart pages.
Drop stale PORT=1852 and JWT_SECRET=your-secure-jwt-secret-here
lines from .env.example so the example no longer contradicts the
docs (PORT is hardcoded, JWT_SECRET is auto-generated and persisted
to the database during initial setup).
* docs(configuration): replace em-dash-substitute hyphens in prose
Three sentences used ` - ` (space-hyphen-space) as an em-dash
substitute. Replaced with the punctuation that fits each case:
- "How Sencho organizes your compose directory": semicolon between
the two related clauses.
- Data directory Warning: parentheses around the parenthetical
insertion.
- 1:1 path rule explanation: comma before the contrastive clause.
Heading slugs and YAML code-block comments are unchanged.
* docs(sso): refresh setup guide and drop misleading "one-click" wording
Brings the SSO Setup Guide in line with how SSO actually works in the
current build, and corrects misleading copy across both SSO docs pages.
- Setup Guide: explains the env-var-seeds-once / DB-is-authoritative
config model up front, replacing per-section "Restart Sencho"
wording that implied a restart was always required.
- Setup Guide: promotes the per-provider Test Connection button out
of the LDAP-only paragraph into a generic intro callout, and adds
a self-signed LDAPS tip.
- Setup Guide: notes that all OIDC providers accept a *_DISPLAY_NAME
override for the login button label, and adds a commented LDAP TLS
toggle to the full compose example.
- Setup Guide: adds two screenshots of the redesigned Settings > SSO
panel (overview + LDAP card expanded with form).
- Both pages: replaces "one-click presets" / "one-click configuration"
with "preset providers". The Skipper-tier presets still require an
OAuth app provisioned in the provider's console; what they actually
buy is provider-aware defaults and a branded login button. The old
wording overpromised.
* docs(features-overview): regroup catalog and add 17 missing features
Restructure the Features Overview into the same six groups the docs
sidebar uses (Stacks & Deployments, Observability, Fleet & Multi-Node,
Security & Identity, Automation, Platform) plus a short Reference tail.
Add catalog entries for 17 shipped features that the previous overview
never mentioned: Stack Activity, Stack File Explorer, Deploy Progress,
Deploy Enforcement, Blueprints, Git Sources, Global Search, Pilot Agent,
Sencho Mesh, Fleet Federation, Fleet Actions, Fleet Sync, Fleet Secrets,
Two-Factor Authentication, CVE Suppressions, Auto-Heal Policies, Stack
Sidebar.
Fix two factual inaccuracies:
- Fleet View blurb wrongly gated search, sort, filter, and stack
drill-down behind Skipper. The deep-dive is explicit that those are
available on every tier; only the bulk Update All action inside the
Node Updates modal is paid.
- Auto-update entry was titled and described as a scheduling system.
The deep-dive page is the Auto-Update Readiness board (risk tags,
changelog previews, rollback targets); the scheduler lives under
Scheduled Operations.
Add three hero screenshots captured from production at 1920x900,
illustrating the redesigned cockpit visual language: Home dashboard,
stack anatomy, and fleet topology.
* docs(stack-management): refresh page around v0.72.0 cockpit and add screenshots
Updates the Stack Management page to match the current UI: sidebar with
filter chips and label groups, bulk mode, restructured kebab menu,
two-tab anatomy panel, and three-source New stack dialog (Empty, From
Git, From Docker Run).
Adds sections for Filter chips, Pinned and label groups, and Bulk mode.
Restructures the Stack context menu around the inspect, organize,
lifecycle, and destructive groups with their keyboard shortcuts.
Documents the From Git tab and cross-links Git Sources for the full
sync flow.
Replaces every existing screenshot with fresh captures from the current
UI and adds eight new captures for the new sections. Cross-links Stack
Activity, Stack Labels, Stack File Explorer, Compose Editor, Atomic
Deployments, Scheduled Operations, Auto-Heal Policies, Auto-Update
Policies, and Alerts and Notifications for features documented on
their own pages.
* docs(stack-activity): refresh page around v1 cockpit and recapture screenshots
Realign the page with the current Anatomy panel tab strip and the
StackActivityTimeline component:
- Frame the Activity tab as a sibling of Anatomy under the right-hand
panel, with files/edit actions belonging to the strip.
- Expand the category guidance: list the five iconized categories and
call out that other stack-scoped notifications (deploy failure,
available image updates, auto-heal triggers, monitor alerts, scan
findings) flow into the timeline with a generic icon.
- Tighten the actor-attribution rule to match the component: omitted
for events without an actor and for system-driven events.
- Add the day-format example to the relative-time row.
- Recapture both screenshots from a populated stack (Today + Yesterday
+ Earlier with three distinct icons) and an empty stack.
- Convert troubleshooting blurbs to H3 for anchor links and consistency
with the v1-refresh sibling pages.
* docs(stack-activity): wrap troubleshooting entries in Accordion blocks
Match the foldable troubleshooting pattern established by
docs/features/deploy-progress.mdx so the page stays compact and
readers can scan to their issue. This is the canonical formatting
for the /features section's troubleshooting blurbs going forward.
* docs(editor): refresh page around v1 cockpit and recapture screenshots
Rewrites the page around the dual-mode right panel (Anatomy by default,
Monaco when the user clicks edit), the redesigned Command Center action
bar, the new container row layout with status badges and live stats, the
Structured / Raw terminal logs toggle, the Git Source toolbar button, and
the opt-in diff preview. Drops the obsolete persistent embedded terminal
section. Preserves the #diff-preview-before-save and #log-viewer anchors
referenced from settings.mdx and global-observability.mdx. Replaces
editor-overview.png and container-exec-modal.png with fresh captures
against v0.72.0 production at 1920x900 dark theme; renames container
-actions.png to containers-list.png; adds command-center.png and
editor-edit-mode.png. PII scrubbed (host paths normalized) and the
compose-diff-preview/diff-modal.png shared asset is left untouched after
visual diff against the live UI showed no chrome change.
* docs(features): wrap troubleshooting accordions in AccordionGroup
Wraps the loose <Accordion> blocks in editor.mdx and stack-activity.mdx
in a single <AccordionGroup> to match the troubleshooting design used
by fleet-federation.mdx. Structural only; no content changes.
* docs(stack-file-explorer): refresh page around v1 explorer and add screenshots
Full rewrite to match the live two-pane Files tab. Replaces the false
'Edit' toolbar flow with the read-only chip + always-on Save model,
fixes the protected-files list (5 names) and dedicated-tab redirect
list (3 names), drops the fabricated 100 MB download cap, and corrects
the upload claim to single file at a time.
Adds sections for New File, Rename, Permissions, the type-to-confirm
protected-file delete flow, the non-empty folder delete confirmation,
the 500-entry tree display cap, and the symlink rendering. Restructures
troubleshooting around <AccordionGroup> + <Accordion> to match
fleet-federation.mdx, and adds two new entries (display cap, 403 on
Community admin write).
Ships nine screenshots captured against v0.72.0 production at 1920x900
in dark theme: overview, two-pane layout, protected-tree-marker,
viewer-edit-mode, new-file-dialog, context-menu-folder, context-menu-file,
permissions-dialog, delete-protected-confirm.
* docs(deploy-progress): rewrite around current modal + capture v1 screenshots
Aligns the page with the v0.72.0 implementation and standardizes the
troubleshooting layout with the rest of the docs refresh.
Setting and gating
- Renames the Settings field to "Deploy progress modal" and quotes the
current helper text verbatim. Documents that the toggle is off by
default, lives under Settings > Appearance > Display, is stored in
localStorage, and syncs across tabs in the same browser.
Modal anatomy
- Names every visible UI string: header verbs, status indicator (with
the "closes in <n>s" countdown that was previously undocumented),
empty-body strings, footer toggle that flips between "Raw output" and
"Hide raw", and the destructive border on ERR rows vs the softer warn
tint on WARN rows.
- Replaces the vague "after a few seconds" with the actual 4-second
auto-close timer; documents hover-to-pause and the
leave-hover-restarts-the-countdown behavior.
- Documents the truncated error message in the failed-state header and
the manual close-only requirement.
- Notes that the pill is portal-mounted and survives navigation.
Stage badges
- Keeps the 9-badge table but adds an honest note that most lines render
as LOG because the badges are gated on Compose's "[+]" progress
prefix, which Compose only emits in TTY mode and Sencho spawns it
without one.
Entry points
- Splits the supported actions into the four that produce a populated
structured-log body (Deploy, Update, Install, Git Apply) and the two
that bypass compose and finish with 0 lines (Restart, Stop). Drops the
"Down" claim from the user-facing list since no UI control currently
triggers it; mentions the down route as an automation surface only.
Troubleshooting
- Wraps the existing accordions in an <AccordionGroup> matching the
pattern used by the editor and stack-activity refreshes. Adds two new
entries: one explaining the Restart/Stop "0 lines" outcome, one
explaining the LOG-everywhere case for non-TTY compose output.
Screenshots (six PNGs in docs/images/deploy-progress/, 1920x900, dark
theme, captured against the upgraded production node)
- setting-toggle.png: the Display section with the toggle enabled.
- modal-streaming.png: a real update in flight at 19s, 554 lines.
- modal-succeeded.png: succeeded state with the live closes-in
countdown visible.
- modal-raw-output.png: structured rows with the Raw output panel
expanded beneath.
- pill.png: minimized pill anchored bottom-center on a stack editor
view.
- modal-failed.png: failed state with the truncated error in the
header and ERR rows highlighted.
* docs(resources): refresh Resources Hub page for v1 redesign and feature additions
Rewrite the page to match the shipping UI and replace stale screenshots
with fresh captures of the redesigned chrome.
- Document the admin-only Reclaim hero and clarify the per-tile Sencho-only
vs. All Docker (includes external) split in Quick Clean.
- Add coverage of the Scan history toolbar button and the per-row severity
badge plus shield-icon scan dropdown in the Images tab; cross-link to the
vulnerability scanning page.
- Correct the Volumes column list (no Size column; size lives on the Largest
5 landing card) and call out admin gates on browse and delete.
- Spell out the List/Topology view-mode toggle and that Create Network is
admin-only and List-mode only.
- Rewrite the Unmanaged tab section around the project-grouped layout, the
Select all + Purge Selected (N) admin-only multi-select, and the empty
state copy.
- Replace screenshots: resources-reclaim, networks-list, create-network,
network-inspect, network-topology, network-topology-toggle. Add fresh
resources-volumes-tab and resources-unmanaged-tab captures.
* docs(app-store): refresh page around v1 deploy sheet, scan integration, and registry settings
Rewrites the App Store reference to match the current cockpit:
- Documents the weekly-rotated featured banner picked from the top-5 by GitHub stars and the star-descending grid sort.
- Adds the deploy-sheet structure (breadcrumb, meta line, About panel with Read more) and splits the Advanced tab into Ports, Volumes, Environment variables, Custom variables, and Security subsections.
- Documents the Trivy-gated Security checkbox, atomic vs non-atomic deploys by tier, and the rollback semantics driven by error class.
- Adds a Watching the deploy section linking to the deploy-progress modal.
- Rewrites the Custom registry section against the new two-panel settings layout (Default + Custom) with the URL validation rule and the using default / using custom hint.
- Adds a four-entry Accordion troubleshooting block in the house style.
- Replaces three screenshots and adds two (Advanced tab, Settings registry panel) captured against the production node.
Permissions wording aligns with current backend (admin only); the broader stack:create gate will land in a follow-up fix branch.
* docs(app-store): describe inline port-conflict messaging on deploy sheet
Update the deploy-sheet section to match the visible port-conflict
behavior: the Essentials tab surfaces a Port-conflict warning that
replaces the defaults hint when any default port is already bound, and
the Advanced tab shows an inline "in use by {stack}" message next to
the container port instead of a hover-only tooltip. Refresh the
screenshot alt-text and the troubleshooting Accordion to match.
* docs(app-store): align permissions note with stack:create gate
Pairs with the backend gate swap in fix/templates-deploy-rbac (#986)
which moves POST /api/templates/deploy from requireAdmin to
requirePermission('stack:create'). Updates the Note block under
'Watching the deploy' so the docs match the new behavior: admin and
node-admin can deploy templates from the App Store; viewer, deployer,
and auditor cannot.
158 lines
13 KiB
Plaintext
158 lines
13 KiB
Plaintext
---
|
||
title: App Store
|
||
description: Browse a curated catalogue of Docker Compose templates and deploy any of them as a new stack with one click.
|
||
---
|
||
|
||
The **App Store** is a browseable catalogue of Docker Compose templates. Each template is a turn-key recipe with sensible defaults for ports, volumes, and environment variables, so you can stand up a new stack without writing YAML. Pick a tile, fill in a name, click **Deploy** and Sencho creates the directory, generates `compose.yaml` and `.env`, runs `docker compose up -d`, and switches you to the editor for the new stack.
|
||
|
||
<Frame>
|
||
<img src="/images/app-store/app-store-hero.png" alt="App Store with the categories rail on the left, a featured template hero pinned at the top, and a grid of editorial tiles with per-tile scan badges" />
|
||
</Frame>
|
||
|
||
## Browsing templates
|
||
|
||
Templates load from a remote registry. The default registry is **LinuxServer.io**; you can point Sencho at any [Portainer v2 compatible JSON catalogue](#custom-template-registry) instead.
|
||
|
||
### Search and counts
|
||
|
||
The search bar at the top filters tiles by name, description, or category in real time. The chip on the right shows the number of templates currently visible (for example, `201 apps`).
|
||
|
||
### Categories rail
|
||
|
||
The left rail lists every category the registry exposes, plus an **All** entry, with a count for each. The active category gets a brand-coloured rail and tinted background. The rail appears once at least one template has loaded.
|
||
|
||
### Featured template
|
||
|
||
A single template is pinned to the top of the grid in a wide editorial banner. Sencho rotates the featured pick **weekly** through the five most-starred templates in the registry, so the catalogue stays fresh without you doing anything. The banner disappears whenever you have a search query active.
|
||
|
||
### Tiles
|
||
|
||
Each tile shows:
|
||
|
||
- A brand-tinted logo (falls back to a generic icon when the registry has no artwork)
|
||
- The template name and the first sentence of its description
|
||
- A GitHub star count (when the registry provides one)
|
||
- The primary category
|
||
- A **scan badge** on the right, derived from Sencho's vulnerability scanner:
|
||
- **Clean** (no CVEs detected on the most recent scan)
|
||
- **{N} CVE** or **{N} CVEs** (the count of open vulnerabilities, in amber)
|
||
- **Unscanned** (no scan has been recorded for this image yet)
|
||
|
||
Tiles are sorted by GitHub star count, descending. Click any tile to open the deployment sheet on the right.
|
||
|
||
## Deploying a template
|
||
|
||
The deployment sheet slides in from the right. The header carries a breadcrumb (`App store › <template>`), the template name in the cockpit display face, and a meta line that combines the supported architectures, the GitHub star count, and the target node name when you are deploying to a remote node.
|
||
|
||
### About
|
||
|
||
The first section is a description block with the template logo, the full description, a **Read more** / **Read less** toggle, and a row of secondary metadata: architecture chips (for example `amd64`, `arm64`), GitHub stars, a **Source** link to the template's repository, and a **Docs** link to the project's official documentation when the registry provides one.
|
||
|
||
### Essentials tab
|
||
|
||
The **Essentials** tab is intentionally minimal: a single **Stack Name** field and a hint that the template's defaults will be used as-is. Click **Deploy {template}** to ship it.
|
||
|
||
<Frame>
|
||
<img src="/images/app-store/deploy-sheet-essentials.png" alt="Deployment sheet open on the Essentials tab, showing the App store breadcrumb, the template hero, an About panel with Read more, the Stack Name field pre-filled with the template slug, and the Deploy with defaults hint card" />
|
||
</Frame>
|
||
|
||
The stack name pre-fills with the template slug (lowercased, hyphens, no spaces). You can change it; the same [naming rules](/features/stack-management#creating-a-stack) apply.
|
||
|
||
If any of the template's default ports are already bound on the active node, the Essentials hint card is replaced by a **Port conflict** warning that lists each conflicting port and the application using it, with a pulsating amber accent. Switch to **Advanced** to remap the host port before deploying.
|
||
|
||
For anything beyond the defaults, switch to **Advanced**.
|
||
|
||
### Advanced tab
|
||
|
||
The **Advanced** tab exposes every knob the template declares. Every field is pre-populated with the template's recommended value, so customisation is additive rather than required.
|
||
|
||
<Frame>
|
||
<img src="/images/app-store/deploy-sheet-advanced.png" alt="Deployment sheet on the Advanced tab for radarr, showing Ports with a host port input, Volumes with three container mount points and editable host paths, and Environment Variables with PUID and PGID defaults" />
|
||
</Frame>
|
||
|
||
#### Ports
|
||
|
||
Each declared port shows an editable **host port** on the left and the fixed **container port** on the right. Host ports must be integers between 1 and 65535; invalid values are highlighted and block deployment.
|
||
|
||
When the host port you typed is already in use somewhere on the active node, an inline amber **`in use by {stack}`** message appears next to the container port, alongside a pulsating amber dot. The message names the offending stack when Sencho manages it, or reads `in use by an external app` when the binding belongs to a process outside Sencho.
|
||
|
||
<Frame>
|
||
<img src="/images/app-store/app-store-port-conflict.png" alt="Port conflict indicator on the radarr deploy sheet: an inline amber message names the stack already using port 7878, next to a pulsating warning dot" />
|
||
</Frame>
|
||
|
||
#### Volumes
|
||
|
||
For each container mount point the template declares, type the **host path** where that data should live. Sencho pre-fills sensible defaults like `./config` or `./data`.
|
||
|
||
<Note>
|
||
Relative paths such as `./config` resolve against the new stack's directory inside `COMPOSE_DIR`. Absolute paths map directly onto the host filesystem.
|
||
</Note>
|
||
|
||
#### Environment variables
|
||
|
||
Each template declares the variables it expects. Sencho pre-fills sensible defaults where it can, and adds three baseline LinuxServer.io conventions automatically when the template does not already define them: **PUID** (`1000`), **PGID** (`1000`), and **TZ**, seeded from your browser's reported timezone.
|
||
|
||
#### Custom variables
|
||
|
||
The **Custom variables** block lets you add arbitrary key/value pairs that the template did not declare. Type a key and value, then click **+** to add it. If you add a key that already exists in the template defaults, a toast warns you that your custom value will override the default (`"{KEY}" already exists in template defaults. The custom value will override it.`).
|
||
|
||
#### Security
|
||
|
||
When [vulnerability scanning](/features/vulnerability-scanning) is configured on the active node, a **Security** section appears at the bottom of the Advanced tab with a single checkbox: **Scan images for vulnerabilities after deploy**. The checkbox is on by default. Leave it on to queue a Trivy scan against every image the new stack pulls. The scan runs in the background once the stack is up; results land on the per-image severity badges in the [Resources Hub](/features/resources) and open the [scan results drawer](/features/vulnerability-scanning#the-scan-results-drawer) when clicked.
|
||
|
||
The Security section is hidden entirely when Trivy is not available on the node.
|
||
|
||
## What happens on Deploy
|
||
|
||
Clicking **Deploy** runs the following sequence:
|
||
|
||
1. **Pre-flight name check.** Sencho asks the active node whether a stack of that name already exists. If it does, you get an immediate error toast and nothing else happens.
|
||
2. **Directory creation.** A new directory is created under `COMPOSE_DIR/<stack-name>` on the active node.
|
||
3. **File generation.** Sencho writes `compose.yaml` (rendered from the template, with your port and volume overrides) and a `.env` file (when you configured environment variables).
|
||
4. **Policy gate.** Any [deploy-enforcement policies](/features/deploy-enforcement) configured on the node run against the generated compose. If a rule blocks the deploy, the directory is cleaned up and you get the rule's reason.
|
||
5. **`docker compose up -d`.** On Skipper and Admiral the deploy is **atomic**: any container that fails to start triggers an automatic rollback to the previous state. On Community the deploy is non-atomic.
|
||
6. **Outcome.**
|
||
- On success, you are switched to the editor for the new stack and a success toast confirms the deploy.
|
||
- On failure, Sencho parses the error. Most failures (image pull, port collision, volume permission, compose validation) trigger a clean rollback: the stack is brought down and the directory is removed. A small set of failures that point to live containers (for example, a startup that crashes after the container is running) leave the stack on disk so you can inspect it. The error toast tells you which.
|
||
|
||
## Watching the deploy
|
||
|
||
Sencho can stream the live `docker compose` output for every template install. Open **Settings › Appearance › Display** and enable **Deploy progress modal**; from then on, every App Store deploy opens a structured log modal you can minimise to a pill that follows you across navigation. See [Deploy Progress Modal](/features/deploy-progress) for the full reference.
|
||
|
||
<Note>
|
||
Deploying a template requires the `stack:create` permission on the target node, which **admin** and **node-admin** roles hold by default. The Deploy button stays disabled when your account does not have it.
|
||
</Note>
|
||
|
||
## Custom template registry
|
||
|
||
By default Sencho uses LinuxServer.io's hosted catalogue. To point at your own:
|
||
|
||
<Frame>
|
||
<img src="/images/app-store/app-store-settings-registry.png" alt="Settings · Advanced · App Store with two panels: Default registry showing LinuxServer.io and the api.linuxserver.io/api/v1/images URL, and Custom registry with a Registry URL input, a 'using default' hint, and Reset to default and Save & refresh buttons" />
|
||
</Frame>
|
||
|
||
1. Open **Settings › Advanced › App Store**.
|
||
2. Type your catalogue URL into **Registry URL**. The URL must start with `http://` or `https://` and serve a JSON array of template objects in the [Portainer v2 template format](https://documentation.portainer.io/v2.0-be/api/templates/).
|
||
3. Click **Save & refresh**. Sencho persists the new URL and immediately invalidates the template cache so the App Store reloads from the new source.
|
||
|
||
A small chip under the input reads **using custom registry** when an override is set and **using default** when the field is empty. Click **Reset to default** to clear the override and fall back to LinuxServer.io.
|
||
|
||
The default registry display at the top of the page is informational only; you cannot remove the LinuxServer.io fallback, and Sencho uses it whenever the custom field is empty.
|
||
|
||
## Troubleshooting
|
||
|
||
<AccordionGroup>
|
||
<Accordion title="No tiles appear">
|
||
Sencho fetches templates from the configured registry on demand. If the grid stays empty and the loading spinner is gone, the registry returned an empty list or the request failed before parsing. Open **Settings › Advanced › App Store** and confirm the **Registry URL** is reachable from the node (try the URL in a browser tab); a custom catalogue served from a private host will not load if the Sencho container cannot resolve or reach it. Click **Save & refresh** to invalidate the cache and try again. If you switched to a custom registry by mistake, click **Reset to default**.
|
||
</Accordion>
|
||
<Accordion title="The port-conflict warning will not clear">
|
||
The conflict indicator is computed against a snapshot of the node's bound ports taken when you opened the deployment sheet. If you freed the port from another stack while the sheet was open, the warning does not refresh on its own. Close the sheet and reopen it to recompute, or pick a different host port. The message `in use by an external app` means a process outside Sencho holds the binding (often the host's reverse proxy or a system service); changing the host port is usually faster than tracking down the offender.
|
||
</Accordion>
|
||
<Accordion title="Deploy failed but the stack directory remains">
|
||
Sencho rolls back the directory only when the failure points cleanly at template generation, image pull, or first-time container creation. Failures that involve a container that did manage to start (for example, a crashloop after Compose returned success, or a healthcheck that flapped) leave the stack on disk so you can read the logs and decide whether to fix forward. The error toast tells you which path was taken; the response also carries a `rolledBack` flag and a `ruleId` you can search for in the [audit log](/features/audit-log). Use the editor to repair the compose, or delete the stack from the cockpit if you want to retry from scratch.
|
||
</Accordion>
|
||
<Accordion title="The Security checkbox is missing from the Advanced tab">
|
||
The **Scan images for vulnerabilities after deploy** checkbox renders only when the active node reports that Trivy is installed. Configure scanning per the [Vulnerability Scanning](/features/vulnerability-scanning) page, then reopen the deployment sheet. Even with the checkbox missing, the rest of the Advanced tab works exactly the same; you just lose the post-deploy scan trigger.
|
||
</Accordion>
|
||
</AccordionGroup>
|