fix: change default endpoint mode from vip to dnsrr for applications#4522
Open
maickonn wants to merge 29 commits into
Open
fix: change default endpoint mode from vip to dnsrr for applications#4522maickonn wants to merge 29 commits into
maickonn wants to merge 29 commits into
Conversation
The empty-records branch of `main()` returned without calling `process.exit(0)`, leaving the Drizzle Postgres connection pool holding the event loop open. The `migrate-auth-secret` process then hangs indefinitely after printing "No 2FA records found, nothing to migrate." causing the upstream `0.29.3.sh` security migration script (which calls this via `docker exec`) to never reach its final `docker service update` step that mounts the new Docker Secret. Operators end up with the new secret created but the dokploy service still configured with the hardcoded `BETTER_AUTH_SECRET`, while believing the migration completed. Match the success branch a few lines below which already does `process.exit(0)`, and the pattern used in sibling scripts `reset-password.ts` and `reset-2fa.ts`. Closes Dokploy#4392
…ret-exit-on-empty fix(migrate-auth-secret): exit cleanly when there are no 2FA records
Adds an "Import" option to the Create Service dropdown that lets users paste a base64-encoded compose export, preview the template (compose YAML, domains, envs, mounts) before confirming, and create the service only on confirm. Adds a `previewTemplate` tRPC procedure that processes the base64 without touching the DB, with server access validation via session.
…-base64 feat(compose): add import from base64 in create service dropdown
- Updated the GitHub Actions workflow to sync versioning across MCP, CLI, and SDK repositories. - Added steps to bump the version in the SDK repository and regenerate tools from the latest OpenAPI spec. - Improved commit message formatting to include source and release information for all repositories. - Ensured successful synchronization messages for each repository after the version update.
- Introduced a new `readLogs` procedure that allows users to retrieve logs for a specific deployment by providing the deployment ID and an optional tail parameter. - Implemented permission checks to ensure users have access to the requested logs. - Enhanced log retrieval for both cloud and non-cloud environments, utilizing appropriate commands based on the server context. Resolve Dokploy/mcp#14
- Implemented server access validation in deployment procedures to ensure users can only access deployments associated with their active organization. - Added checks to throw an UNAUTHORIZED error if a user attempts to access a deployment linked to a server outside their organization. This enhancement improves security and access control within the deployment management system.
- Added validation to prevent users from being invited with the owner role in the organization and user routers. - Implemented TRPCError responses to ensure proper error handling when attempting to assign the owner role. This change enhances role management and security within the organization structure. https://github.com/Dokploy/dokploy/security/advisories/GHSA-fm9p-wmpw-gxjh
- Added functionality to delete old sessions when a user updates their password, ensuring that only the current session remains active. - This change enhances security by preventing unauthorized access from previous sessions after a password change. Close here https://github.com/Dokploy/dokploy/security/advisories/GHSA-rr9m-w87g-46f3
* fix: copy Dokploy server IP when clicking server badge When a service runs on the local Dokploy server (no remote server), clicking the server badge did nothing because `data.server` is null. Now falls back to the server IP from settings so the badge always copies an IP address. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat(copy-ip): implement IP address copying functionality across database service components - Added the ability to copy the server IP address to the clipboard when clicking the server badge in various database service components (Libsql, MariaDB, MongoDB, MySQL, PostgreSQL, Redis). - Integrated the `copy-to-clipboard` library and `sonner` for user feedback upon successful copy action. - Ensured fallback to the server IP from settings when the service data is not available, enhancing user experience and functionality. --------- Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Co-authored-by: Mauricio Siu <siumauricio@icloud.com>
Signed-off-by: Nahidujjaman Hridoy <hridoyboss12@gmail.com>
…of register, and build for extra. (Dokploy#4382)
… routes (Dokploy#4468) * fix: allow square brackets in zip drop path validation for Next.js dynamic routes ZIP uploads containing Next.js dynamic route files (e.g. app/api/[id]/route.ts, pages/[slug].tsx) were rejected by readValidDirectory because the path regex did not include square bracket characters. * [autofix.ci] apply automated fixes --------- Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
…es (Dokploy#4470) shouldDeploy passed undefined/null entries from commit.modified straight into micromatch, which throws "Expected input to be a string" and fails every webhook deployment when watch paths are configured. Filter out non-string values before matching.
…t accidental submission (Dokploy#4422) Co-authored-by: Maks Pikov <mixelburg@users.noreply.github.com>
…ploy#4018) (Dokploy#4474) * fix: add tls=true label for compose domains when certificateType is none (Dokploy#4018) * test: cover tls=true label for certificateType none, require https * fix: scope tls fix to compose labels, leave traefik file config unchanged (Dokploy#4018)
Upgraded next dependency in apps/dokploy to 16.2.6 exactly. Verified typescript typecheck passes successfully.
…nforce-sso) (Dokploy#4511) * feat: add self-hosted enterprise restrictions (remote-servers-only, enforce-sso) - Add `remoteServersOnly` field to webServerSettings: prevents creating services on the local Dokploy VM, forcing all deployments to remote servers. Validated in all 8 service routers (application, compose, postgres, mysql, mongo, redis, mariadb, libsql). - Add `enforceSSO` field to webServerSettings: hides the email/password login form and shows only the SSO button on the login page. - Both settings are enterprise-only (enterpriseProcedure) and self-hosted-only (blocked at the API level when IS_CLOUD=true). - UI toggles added to the SSO settings page under a new "Self-hosted Restrictions" card (hidden in cloud). Login page reads enforceSSO from getServerSideProps to avoid client-side flash. - Migrations: 0167_fresh_goliath.sql, 0168_long_justice.sql * fix: add missing final newlines to migration files * refactor: improve code formatting for better readability in multiple components - Adjusted formatting in `add-application.tsx`, `add-compose.tsx`, and `add-database.tsx` to enhance readability by adding line breaks and consistent indentation. - Updated `toggle-enforce-sso.tsx` to simplify the Switch component's props. - Reformatted imports in `index.tsx` and `sso.tsx` for consistency. - Cleaned up conditional statements in various router files for improved clarity. * fix: add enforceSSO to test mock
…eys is enabled for members (Dokploy#4512)
On settings/servers, a long server name in the card title (h3) did not wrap and overflowed its container, overlapping nearby content and squeezing the three-dots actions menu until it disappeared. Allow the title block to shrink and wrap (min-w-0 + break-words), keep the server icon and the actions trigger from being crushed (shrink-0), and add gap between the title and the actions button.
… docker config (Dokploy#4485) The compose/stack deploy command runs under `env -i PATH="$PATH"`, which clears the environment except for PATH. That strips HOME, so when the generated command is `docker stack deploy --prune --with-registry-auth` the docker CLI cannot resolve `~/.docker/config.json` (e.g. `/root/.docker/config.json`) and ships no registry credentials to the swarm. Private-registry images then fail to pull on the nodes: image registry.example.com/... could not be accessed on a registry to record its digest. Each node will access ... independently while the deploy still logs "Docker Compose Deployed: ✅". Keep PATH isolation but preserve HOME so docker can read its config for both `stack deploy --with-registry-auth` and `compose up -d --build`. Add a regression test asserting the generated command preserves `HOME="$HOME"` for both stack and docker-compose deploys. Fixes Dokploy#4401 Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
dosubot
Bot
added
the
size:S
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
10 participants
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.
Problem
Applications deployed with Dokploy on Docker Swarm frequently return 502 Bad Gateway when accessed through Traefik. The root cause is that Docker Swarm's default endpoint mode (
vip) relies on IPVS (IP Virtual Server) for internal load balancing, which fails in several common environments:EPERM(confirmed viastrace, dockerd logs show"Failed to create a new service for vip ... operation not permitted")-EPERMeven for root with full capabilitiesDokploy's built-in databases (PostgreSQL, MySQL, MariaDB, Redis, MongoDB, LibSQL) already default to
dnsrrmode and work correctly in all these environments. Applications were the only service type still defaulting tovip.Solution
Change the default endpoint mode for applications from
vip(implicit Docker default) todnsrr(DNS Round Robin), matching what databases already use.packages/server/src/utils/builders/index.tsMode: "dnsrr"to application EndpointSpec fallbackpackages/server/src/services/rollbacks.tsMode: "dnsrr"to rollback EndpointSpecapps/dokploy/.../endpoint-spec-form.tsxdnsrras defaultHow
dnsrrworksInstead of a Virtual IP (VIP) that requires IPVS kernel routing,
dnsrrmakes the Swarm DNS resolver return the container IPs directly. Traefik receives the real container IPs and performs client-side load balancing via its own health checks — which it already does for all backends regardless of endpoint mode.Compatibility
vip: Can still switch via UI (Advanced → Cluster → Swarm Settings → Endpoint Spec).Related issues