a month ago
I deployed the Langfuse v3 container and both the Redis service and the worker failed to start.
Redis Service Failure
The version of redis tagged in the template doesn't exist on dockerhub and can't be pulled. Migrating over to Redis 8.4.0-alpine worked just fine though
Worker Service Failure
The worker service seems to fail because it doesn't have a proper NEXTAUTH_URL variable set and doesn't have a domain generated. That said, I don't see why it would need that if it's a worker- I would assume that it should be internal only on not need a public URL.
Would really love to try out Langfuse on Railway and I'm sure I can spend some time and debug why the worker service is down but the template should probably just be a one-click deploy to a healthy state 
2 Replies
a month ago
Summary of Fixes
1. Redis — Invalid Docker image tag
- Problem: Container failed to start because the Bitnami Redis image (bitnami/redis) doesn't have a 7.2.5 tag. Bitnami uses full immutable tags
like 7.2.5-debian-12-r4 and skipped 7.4 entirely.
- Fix: Changed the image/tag in Railway dashboard (you handled this one)
2. Langfuse Worker — Broken NEXTAUTH_URL
- Problem: NEXTAUTH_URL was set to just https:// — an incomplete URL. Likely a Railway variable reference
(${{langfuse-web.RAILWAY_PUBLIC_DOMAIN}}) that failed to resolve, possibly cascading from the Redis outage. Zod validation rejected it on
startup.
- Fix: Set NEXTAUTH_URL directly to https://langfuse-web-production-b2db.up.railway.app via railway variable set. Worker redeployed and came up
healthy.
miktomic
Summary of Fixes1. Redis — Invalid Docker image tag- Problem: Container failed to start because the Bitnami Redis image (bitnami/redis) doesn't have a 7.2.5 tag. Bitnami uses full immutable tagslike 7.2.5-debian-12-r4 and skipped 7.4 entirely.- Fix: Changed the image/tag in Railway dashboard (you handled this one)2. Langfuse Worker — Broken NEXTAUTH_URL- Problem: NEXTAUTH_URL was set to just https:// — an incomplete URL. Likely a Railway variable reference(${{langfuse-web.RAILWAY_PUBLIC_DOMAIN}}) that failed to resolve, possibly cascading from the Redis outage. Zod validation rejected it onstartup.- Fix: Set NEXTAUTH_URL directly to https://langfuse-web-production-b2db.up.railway.app via railway variable set. Worker redeployed and came uphealthy.
a month ago
Thanks! I was able to get it working just by messing around, but users shouldn't have to do this.
Hoping that Langfuse can update their template so the default experience is solid.