fix(firecrawl): align self-host template with v2.8.0#769
fix(firecrawl): align self-host template with v2.8.0#769Manan-Santoki wants to merge 7 commits intoDokploy:canaryfrom
Conversation
built with Refined Cloudflare Pages Action⚡ Cloudflare Pages Deployment
|
dosubot
bot
added
the
size:L
Tested on my Dokploy instance and the updated template works as expected. Verified:
Attached screenshots from my test instance for reference. Let me know if you want me to verify any other edge cases. |
There was a problem hiding this comment.
Pull request overview
Updates the firecrawl Dokploy blueprint to match upstream Firecrawl v2.8.0’s current self-hosted architecture (harness-driven API + required dependencies), and refreshes the associated metadata entry.
Changes:
- Replaces the legacy split-process Firecrawl services with a single harness-driven
apiservice pinned tov2.8.0. - Adds RabbitMQ as a required dependency and updates env wiring to
POSTGRES_*andNUQ_RABBITMQ_URL. - Updates
meta.jsonFirecrawl version/description and removes a pre-existing duplicatestrapientry during meta processing.
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
| meta.json | Pins Firecrawl metadata to v2.8.0, refreshes description, and removes a duplicate strapi entry. |
| blueprints/firecrawl/template.toml | Updates Dokploy variables/env model for the new upstream service layout (Postgres, RabbitMQ, optional provider/tuning envs). |
| blueprints/firecrawl/docker-compose.yml | Switches to upstream v2.8.0 build contexts and harness entrypoint; adds RabbitMQ + volumes and updates env defaults. |
| POSTGRES_USER: ${POSTGRES_USER:-postgres} | ||
| POSTGRES_PASSWORD: ${POSTGRES_PASSWORD:-firecrawl} | ||
| POSTGRES_DB: ${POSTGRES_DB:-postgres} | ||
| NUQ_RABBITMQ_URL: ${NUQ_RABBITMQ_URL:-amqp://rabbitmq:5672} |
There was a problem hiding this comment.
NUQ_RABBITMQ_URL defaults to amqp://rabbitmq:5672 (no credentials). The RabbitMQ official image’s default guest/guest user is restricted to localhost by default, so the API container typically won’t be able to authenticate over the Docker network and Firecrawl/NuQ will fail to connect. Define a non-guest RabbitMQ user/password and include them in NUQ_RABBITMQ_URL (or build the URL from RABBITMQ_DEFAULT_USER/RABBITMQ_DEFAULT_PASS).
| NUQ_RABBITMQ_URL: ${NUQ_RABBITMQ_URL:-amqp://rabbitmq:5672} | |
| RABBITMQ_DEFAULT_USER: ${RABBITMQ_DEFAULT_USER:-firecrawl} | |
| RABBITMQ_DEFAULT_PASS: ${RABBITMQ_DEFAULT_PASS:-firecrawl} | |
| NUQ_RABBITMQ_URL: ${NUQ_RABBITMQ_URL:-amqp://${RABBITMQ_DEFAULT_USER:-firecrawl}:${RABBITMQ_DEFAULT_PASS:-firecrawl}@rabbitmq:5672} |
| rabbitmq: | ||
| image: rabbitmq:3.13-management | ||
| restart: unless-stopped | ||
| command: rabbitmq-server | ||
| volumes: | ||
| - rabbitmq_data:/var/lib/rabbitmq |
There was a problem hiding this comment.
The rabbitmq service doesn’t configure RABBITMQ_DEFAULT_USER / RABBITMQ_DEFAULT_PASS, which means it will fall back to the default guest account (localhost-only by default in the official image). Add explicit credentials (and optionally a vhost) so other containers can connect reliably, and align them with NUQ_RABBITMQ_URL.
| "POSTGRES_HOST=nuq-postgres", | ||
| "POSTGRES_PORT=5432", | ||
| "POSTGRES_USER=${postgres_user}", | ||
| "POSTGRES_PASSWORD=${postgres_password}", | ||
| "POSTGRES_DB=${postgres_db}", | ||
| "REDIS_URL=redis://redis:6379", | ||
| "REDIS_RATE_LIMIT_URL=redis://redis:6379", | ||
| "NUQ_RABBITMQ_URL=amqp://rabbitmq:5672", | ||
| "PLAYWRIGHT_MICROSERVICE_URL=http://playwright-service:3000/scrape", |
There was a problem hiding this comment.
template.toml hardcodes NUQ_RABBITMQ_URL=amqp://rabbitmq:5672 without credentials. With the default RabbitMQ image settings, this will usually attempt guest/guest and fail from non-localhost clients. Add template variables for RabbitMQ user/password (or generate a secret), set RABBITMQ_DEFAULT_USER/RABBITMQ_DEFAULT_PASS, and update NUQ_RABBITMQ_URL to include them.
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
| rabbitmq: | ||
| image: rabbitmq:3.13-management | ||
| restart: unless-stopped | ||
| command: rabbitmq-server | ||
| volumes: | ||
| - rabbitmq_data:/var/lib/rabbitmq | ||
| healthcheck: | ||
| test: ["CMD", "rabbitmq-diagnostics", "-q", "check_running"] | ||
| interval: 5s | ||
| timeout: 5s | ||
| retries: 3 |
There was a problem hiding this comment.
The default NUQ_RABBITMQ_URL (amqp://rabbitmq:5672) carries no credentials, so the AMQP client falls back to guest/guest. RabbitMQ 3.x blocks the built-in guest account from any non-loopback address by default — a container-to-container connection is never loopback, so the api service will receive ACCESS_REFUSED on startup.
Either expose a dedicated user via the rabbitmq service and include credentials in the URL:
rabbitmq:
image: rabbitmq:3.13-management
restart: unless-stopped
environment:
RABBITMQ_DEFAULT_USER: ${RABBITMQ_DEFAULT_USER:-firecrawl}
RABBITMQ_DEFAULT_PASS: ${RABBITMQ_DEFAULT_PASS:-firecrawl}
...And update the default URL:
NUQ_RABBITMQ_URL: ${NUQ_RABBITMQ_URL:-amqp://firecrawl:firecrawl@rabbitmq:5672}The same change is needed in template.toml line 67.
| BLOCK_MEDIA: ${BLOCK_MEDIA:-} | ||
| NO_PROXY: ${NO_PROXY:-localhost,127.0.0.1,redis,nuq-postgres,playwright-service,host.docker.internal} | ||
|
|
||
| ALLOW_LOCAL_WEBHOOKS: ${ALLOW_LOCAL_WEBHOOKS:-false} |
There was a problem hiding this comment.
ALLOW_LOCAL_WEBHOOKS does not belong in playwright-service
ALLOW_LOCAL_WEBHOOKS is a Firecrawl API-level flag that guards outbound webhook delivery. The playwright browser service has no webhook logic and does not read this variable. Its presence here is harmless but misleading — it creates the impression that removing it from the api service env block would still be covered, which it would not be.
| ALLOW_LOCAL_WEBHOOKS: ${ALLOW_LOCAL_WEBHOOKS:-false} | |
| BLOCK_MEDIA: ${BLOCK_MEDIA:-false} |
Summary
v2.8.0and switch startup to the upstream harness entrypointmeta.jsonentryWhat changed
apiservicerabbitmqdependency and keptredis,playwright-service, andnuq-postgreslatestimages to remote source builds pinned tov2.8.0POSTGRES_*,NUQ_RABBITMQ_URL, and current optional Firecrawl envsv2.8.0strapientry frommeta.jsonRoot cause
The existing Firecrawl template still launched the old runtime layout using
node --import ./dist/src/otel.js .... Upstream Firecrawl now starts throughnode dist/src/harness.js --start-dockerand expects RabbitMQ plus harness-managed workers, so the old Dokploy template crashed withERR_MODULE_NOT_FOUNDfor/app/dist/src/otel.js.Impact
Dokploy users should get a Firecrawl deployment shape that matches current upstream self-hosting instead of the broken legacy image startup path. The template is also now version-pinned instead of following
latest.Validation
node dedupe-and-sort-meta.jsdocker compose -f blueprints/firecrawl/docker-compose.yml confignpm run validate-docker-compose -- --file ../blueprints/firecrawl/docker-compose.yml(frombuild-scripts/)npm run validate-template -- --dir ../blueprints/firecrawl(frombuild-scripts/)git diff --checkGreptile Summary
This PR aligns the Firecrawl blueprint with the v2.8.0 upstream self-hosted architecture, replacing the broken legacy split-process startup with the harness-driven layout, adding RabbitMQ, and pinning all source builds to the
v2.8.0git tag.NUQ_RABBITMQ_URL(amqp://rabbitmq:5672) omits credentials. RabbitMQ's built-inguestaccount is restricted to loopback connections, so theapicontainer will receiveACCESS_REFUSEDand fail to start. The rabbitmq service needs a configured user and the URL needs matching credentials.Confidence Score: 3/5
Not safe to merge — the RabbitMQ URL missing credentials will prevent the api service from connecting to RabbitMQ on startup.
One P1 defect: the default AMQP URL has no credentials and RabbitMQ's guest account is blocked from non-localhost connections, meaning the api container will fail to start in a default Dokploy deployment. The rest of the changes are structurally correct and well-aligned with the upstream v2.8.0 layout.
blueprints/firecrawl/docker-compose.yml (rabbitmq service and NUQ_RABBITMQ_URL default) and blueprints/firecrawl/template.toml (NUQ_RABBITMQ_URL default on line 67)
Reviews (1): Last reviewed commit: "fix(firecrawl): default optional env var..." | Re-trigger Greptile
(3/5) Reply to the agent's comments like "Can you suggest a fix for this @greptileai?" or ask follow-up questions!