Portal provides forms for configuring services. The public deployment API is a separate interface for automation; not every Portal setting has a corresponding public endpoint.
Services#
On your slice, choose Add service, enter its name and type, then configure it. You can create a service before choosing an image. An image is required before it can start.
In service settings:
- Source: choose or edit the deployment source. For container images, enter the image reference. Enable Registry requires login for a private image and supply a username and password or access token. Saved passwords and tokens are not returned to the browser.
- Storage: add a new volume or attach an existing slice volume. Choose the
container and an absolute mount path, such as
/data. A single container is selected automatically. Each mount has its own row, including when the same volume is attached at multiple paths. Edit changes a mount; Remove detaches it without deleting the volume or its files. Mount changes take effect on deployment. - Endpoints: configure the ports your container listens on and their exposure.
- Environment: add the variables and secrets your application requires.
To trust the Budgethost HTTP proxy, add or edit the variable your app expects
(for example,
SIMPLEDATA_TRUST_PROXY) and choose Value source: Platform proxy. Budgethost supplies the trusted proxy range at deployment. Build time and Secret are disabled for this value source. Your application must use the variable to configure proxy trust. - Runtime files: add text files or upload binary files, including images. Set a
relative runtime path such as
branding/logo.pngand an absolute container path such as/app/public/logo.png. Select the container and file permissions.
Runtime files are limited to 128 KiB per file, 512 KiB total per service and 20 files. Limits count decoded bytes, including binary uploads. PNG, JPEG, GIF and WebP uploads have previews; text remains editable. Use persistent storage for larger files and application data.
Save settings to apply the service configuration. For registry services, saving an image configuration pulls the image and recreates the container. Storage attachment changes are saved separately and need a deployment before the mount changes at runtime. The displayed mount status reflects reported runtime observations and may be unavailable until the platform has received them.
Deleting a service removes the service and its components. Storage and backups are retained. Controls remain disabled while deletion finishes.
Change deployment source#
For an existing custom application, open Application settings > Source > Change source. Choose Container image, API upload, or GitHub repository. Managed services such as MariaDB and WordPress do not offer this selector.
Save other settings first. Save source preserves the service ID, runtime template, ports, health checks, domains, storage, environment and secret bindings. It does not queue a deployment or restart the running application. The next image must work with those existing runtime settings.
- Container image: supply the image reference and, for a private registry, its login. Then use Redeploy now to deploy it.
- API upload: use a service deployment token to upload source or a saved image through the existing deployment API. Token permissions and readiness checks still apply.
- GitHub repository: enter the repository, ref and application directory. Choose a Repository Dockerfile in that directory, or the existing Node.js commands build method. Private repositories need an access token. Then use Rebuild now.
Saving a source disables automatic GitHub release deployment. Enable it again under Deployment after reviewing the new source. API deployment creation and API configuration writes are rejected while an explicitly selected source is not API upload. Existing tokens are not revoked. Changing repositories does not reuse the old repository's credential; registry logins must match the selected registry.
Source changes require application management access and are blocked while deployment
or other protected service work is active. Reload settings if another user changed
them after you opened the dialog. Repository Dockerfile builds require an updated build
agent advertising git_dockerfile_build_v1.
The same source setup is available through the public deployment settings API,
using a service token with the opt-in deployment:configure permission. Enabling
automatic releases and triggering a deployment also require deploy.
Build allowances#
Application settings show the service's manual-build allowance and GitHub automatic-build allowance. Defaults are 12 manual builds and 6 automatic builds per rolling 24 hours. An unlimited allowance is displayed as Unlimited, with no remaining-build count.
You can view allowances in Portal; contact support to request changes. Manual rebuilds and automatic builds each have a separate five-minute cooldown, including when the daily allowance is unlimited. Cooldowns start when a build is accepted. The manual allowance also applies to registry application redeploys. If an allowance is lowered, earlier accepted builds still count until they leave the rolling 24-hour window.
External domains#
In Domains & Routes, choose Add domain, then External domain. Enter the root
domain (for example, example.com), add the displayed TXT record at your current
DNS provider, then choose Check verification. If the record is not visible
yet, keep the dialog open and try again after DNS has updated. The verification
expires after 24 hours; use Start again if it expires.
Once connected, choose Add route to connect a hostname such as
app.example.com to a service. Publish that hostname's DNS records at your
existing provider. Connecting the domain leaves its registrar, nameservers and
existing DNS records unchanged.
The flow requires workspace management access. Domains already assigned to another workspace require help from support.
An external domain's card also has Transfer to Budgethost. It opens the transfer order with the domain name filled in. Review the price and registered owner before requesting the transfer. Existing nameservers stay in place by default; a DNS switch requires a separate review and confirmation. An active transfer disables the button.
Trust forwarded client addresses#
In application settings, add or edit the environment variable your application
uses for proxy trust and choose Platform proxy as the value source. Use the
variable name expected by your application, then save the settings. Your
application must read that variable; custom images are not changed automatically.
PLATFORM_PROXY_UNAVAILABLE means the required configuration is unavailable;
contact support before retrying.
Managed WordPress, static Nginx, Webmail and phpMyAdmin services receive their proxy-trust configuration automatically when applied. Changes take effect on the next service deployment.
Customer DNS templates#
The page separates Budgethost templates from My templates. Budgethost templates are available in every workspace. Customers can apply them directly or choose Duplicate to create an independent, editable workspace template. Budgethost templates are read-only. Linked-domain counts and update status show only your workspace.
The initial preset, Parked domain - no website or email, contains a root
null MX (0 .), root SPF (v=spf1 -all) and _dmarc TXT
(v=DMARC1; p=reject; sp=reject; adkim=s; aspf=s), all with TTL 3600. It has no
website or service address records. Before applying it, Budgethost checks enabled
service routes, manual routing rules and mail domains belonging to the workspace,
including subdomains. The preview lists any services that prevent the change.
Disconnect those services before applying the template. If usage changes after
your review, open a new preview and resolve the reported conflicts.
Domains linked to a Budgethost template follow its updates. Duplicated personal templates remain independent, so you control when their records change.
In Domains & Routes > DNS templates, create a template with a name and DNS
records. Names are relative: @ is the root, www is a subdomain, and
_autodiscover._tcp is an SRV owner. Supported customer record types are A, AAAA,
ALIAS, CNAME, MX, SRV and TXT. MX values use priority hostname; SRV values use
priority weight port hostname. TTLs are between 60 and 2419200 seconds.
On a domain using Budgethost managed DNS, choose Apply template, review the additions and removals, and confirm the replacement. Every editable record that is absent from the template is removed. Applying an empty template removes all editable records. Provider defaults and record types outside the customer editor are preserved, as are nameservers and DNSSEC settings. A conflicting DNS target used by a route blocks the update; keep that target in the template or remove the route first. Disconnect an existing DNS template before applying another one.
When moving a domain to Budgethost DNS, choose a Budgethost or personal template under DNS settings for this move. Review its records and confirm the replacement before switching nameservers. This uses the complete template and preserves discovered unmanaged records; choose Copy current DNS records to customize an imported selection instead. The domain links to the template when migration completes and follows subsequent updates. Empty templates cannot be used for a DNS move. Parked templates require active routes and mail services to be disconnected before approval.
The domain remains linked to the customer template. Saving changed template records requires a review of all linked domains and queues their updates. Each domain reports Updating, Up to date or Needs attention. Failed updates retry with backoff; Retry updates schedules an immediate retry. Disconnect retains the current DNS records and enables individual record editing. A linked template cannot be deleted. Personal templates and domain operations are scoped to the current workspace.
DNS template changes require workspace management access and may ask you to verify your session. If a review expires because the template or DNS records changed, open a fresh preview and confirm the updated changes.
Rewrite routes#
A rewrite matches its source path exactly and sends that request to the selected
service and target path. For a rewrite whose source path is /, enable Forward
other paths to send other paths to the same service without changing their URI.
For example, / can open /stream, while /styles.css and /images/logo.png
continue to reach those paths on the service. Query strings are preserved.
The option defaults to off. Existing rewrites keep their exact-path behavior.
The service's path, user-agent and IP filters apply to both the rewritten request
and other forwarded paths. Forwarding other paths requires source path /.
For an application served under a subdirectory, choose Mount a subpath instead.
With source / and target /admin, / goes to /admin/, /styles.css goes to
/admin/styles.css, and query strings remain intact. Mount mode requires
source / and Forward other paths turned off. Its target must be a plain absolute
path without a query, fragment, percent encoding, repeated slashes, or dot
segments. These options are mutually exclusive; mounting changes every path,
while forwarding keeps other paths unchanged. Full-app routes reserve their
hostname so another route cannot silently override their mapping.
Service filters and ownership restrictions also apply to mounted requests. Applications must support serving from the public root; proxy mapping cannot rewrite absolute paths embedded in HTML or JavaScript.
DNS migration reviews#
DNS migration previews show warnings and any issues that block the move. When applying a Budgethost DNS template, record lookup failures are advisory because the template provides the replacement records. Only discovered unmanaged records can be preserved. Customers review that limitation in the template confirmation. Copying existing records still requires successful discovery. Delegation and DNSSEC inconsistencies must be resolved before confirming either kind of move.
Domain rows show DNS transfer while migration is active, including propagation and DNSSEC restoration. The hover text gives the current phase. Paused operations show DNS transfer paused or DNS review needed and appear in Issues only.
Slice settings#
Storage lists the slice's volumes, including unattached volumes retained after a service was deleted. Attach connects a volume to a service. Delete permanently removes a volume and its files; attached volumes cannot be deleted, and snapshots must be removed first.
API tokens lists both slice and service tokens. Create a named token with the scope and permissions it needs. Expand its row to see its ID, permissions and scope, or rename it. Renew opens a dialog containing the replacement credentials. Update your clients before revoking the old token. Delete is available after revocation.
Choose Permissions on an active token to change its access, including Configure deployment source and builds for GitHub deployment setup. Review the additions and removals, confirm any added access, then choose Save permissions. Changes apply to everyone holding that token; its secret, scope and expiry stay the same. Other tokens keep their permissions. Expired and revoked tokens cannot be edited. If another administrator changes the token while the editor is open, close and reopen it to review the latest permissions before saving again. Editing requires workspace management access and the workspace's credential security policy, just like creating or revoking a token.
Slice settings refreshes automatically while open.
Import data into a service#
On Services, find the destination slice and choose Import data. This uses the slice's databases and attached writable storage; no backup schedule or existing recovery point is required. If no destinations are listed, first add a database or attach storage to a service in that slice.
Select the destination and upload a database export (.sql.gz) or service-file
archive (.zip). Confirm stopping slice services and pausing external writers, then choose Prepare and validate to inspect it. The archive's
content folder is detected automatically where possible; select a folder if the
preview asks you to resolve an ambiguous layout. Review the destination and path
mapping, confirm the empty destination, then choose Import into empty destination.
To replace existing data, choose Clear and import before validation. Review the destination and existing contents, approve deletion, and type the exact target ID before choosing Clear destination and import. Keep a separate backup; a failed replacement requires manual recovery.
Imports use the backup restore lifecycle to stop writers and run the selected database normally. Imports never merge data. Applications remain stopped by default; optionally select Resume during review to restart only previously running application services after verification. Failures require recovery and do not trigger application restart. Database imports require an already-deployed database. Import files and databases separately when cloning an application.
The entry point requires workspace management and upload access. Validation and applying the import additionally require import access. These are separate from backup scheduling permissions.
Portal and public API differences#
Portal supports text/binary runtime files and token renaming through its forms. Use only the operations listed in the public API reference for automation. Registry credentials can also be managed through the registry login API. Service-token creation through the slice API accepts a token name.
The public deployment API accepts Dockerfile source archives and docker save
image archives. Its configuration flow uses a manifest, a preview, and an apply
request. Use the documented API schemas rather than translating Portal form fields
into guessed API fields.
Add a managed service#
On Services, choose Add managed service in the destination slice. MariaDB is currently the only available type. Enter the service name and logical database name, create it, then review and choose Deploy service. Deploy later leaves the service configured; you can reopen the dialog to continue. No existing application or data is converted. See Managed services for permissions, private networking, progress and retry semantics.