Hey there! We've found the following might help you get unblocked faster:

• 🧵 URGENT - Database corruption after failed n8n upgrade

• 🧵 n8n → Postgres connection issue: “Role ‘postgres’ does not exist

• 🧵 URGENT: PostgreSQL Service Fails to Start with Existing Volume - n8n Deployment Down Due to Authentication Mismatch

If you find the answer from one of these, please let us know by solving the thread!

Hi I need help urgently. My projects are all offed due to the crash.

I do not understand how the official can "automatically" help me upgrade any of my services, especially when it has serious compatibility issue.

Let me provide more information. It appears to have a collation version mismatch.

Also, Postgres is also having trouble to connect with database. Under Postgres - Database, it shows:
- OK with "Deployment Online"
- OK with "Required Variables"
- Forever loading for "Database Connection"

Below is the deployment of the Postgres log:

2025-09-09 15:45:36.903 UTC [124] WARNING: database "railway" has a collation version mismatch

2025-09-09 15:45:36.903 UTC [124] DETAIL: The database was created using collation version 2.36, but the operating system provides version 2.41.

I am unable to post more error log, as it appears that the it may contains some confidential information and the forum disallows it.

i have the same problem, no answer

I am seeing a lot of the users are facing the same issue, and some of the threads have at least been responded by a railway employee but not mine. It has been more than 10 hours since I stated this thread but still do "awaiting railway response".

I tried proyectos2025ia's method in another threads to redeploy "Primary" and "Postgres" and I can login to my n8n for now but every error in Postgres is still showing the same. I also see proyectos2025ia is reporting another crash later.

I suggest everyone who tried the proyectos2025ia's method to login to their n8n and download all the workflow JSON as a safe-net plan, and await for official explanation on what is happening. I am still convinced that it is due to the automatic update with "worker" on railway metal as I "witnessed" the crash time when I am setting up my workflow in n8n, the time of the crash matches with the automatic update of railway metal.

OMG. Unable to even login now. This is crazy!
Looks like self hosting on railway is a bad idea.

Does anyone have links to deployments that are failing? It is hard for us to look into this without specific deployments or logs. Also, please ensure that n8n Primary and n8n Worker services have been redeployed and are on the same version. If the versions are mismatched then it can cause issues.

How to provide the logs to you? What logs do we need here. Currently, my n8n is login-able but Postgres deployment logs are full of red (errors). Thanks.

Link to the Railway deployment page.

Postgres logs everything to stderr, even if it is not an error. The collation mismatch will not cause issues with Postgres and just appears as a warning. But since you only did a minor version bump then it is okay.

Do you mean I post my link to railway deployment page here?

Yes please.

Can you re-iterate the issues you're having? The collation mismatch does not impact Postgres, and Postgres logs show up as error because Postgres logs to STDERR by default which we have no control over. Are you having issues connecting to this Postgres?

Originally, on September 9, I was working on my n8n project when it suddenly stopped working. I tried refreshing the page, but instead of reopening my project it redirected me to a login page — which was unusual. When I entered my credentials, I received the error: "Problem logging in column User.role does not exist".

I then went to my Railway project page to check for issues. All four modules (Primary for n8n, Worker for n8n, Postgres, and Redis) appeared active. However, I noticed something odd: my n8n Worker module showed as having been deployed only a few minutes earlier, while my Primary module had last been deployed a week or more ago. Normally, I always sync deployments of both Primary and Worker, so this was unexpected. Looking deeper into the Worker module logs, I found a line I had never seen before — something about an automatic upgrade for Railway Metal, with a star icon or something similar beside it. (Unfortunately, I didn’t capture this line, as I was panicked and frustrated at the time of the crash.)I immediately opened this thread to seek official support. I also have deployment logs from Postgres (which are full of errors) and from the Worker module (the one that appears to have been auto‑deployed). I wasn’t allowed to post them here — is there a way I can share them directly with the Railway team, perhaps through email or private message?

On September 10, I noticed that some of my automations began partially working — I received a few emails triggered by workflows, but not all of them went through. When I tried logging in again, the same login error remained.

I checked other discussion threads and still didn’t see an official response, so I kept reading. I found one suggestion to redeploy the Primary, Worker, and Postgres modules. I tried that, and I was finally able to log into n8n again.

At the moment, n8n seems to be working normally. However, I still see ongoing issues in the logs:

• Postgres module: The “Database Connection Status” still shows as loading instead of a checkmark. I don’t recall whether this was ever a checkmark when things were working fine before, but it has stayed in the “loading” state since the crash. The deployment log also includes collation mismatch related errors.

• Redis module: Since the crash, I now see a warning under “Required Variables”. I am not sure if this warning was present previously while n8n was working properly.

Please let me know if you need further information. Thank you.

Hi, I updated the database and redis it. I also redid it for Redis, Workers, and Primary in the same way. I didn't get the role error. Could you try it?

Wired stuff is still happening.
There was QUEUE_BULL variables that had to be removed. Followed another thread that an employee answered. It worked for a couple of days.

Today, my webhooks are failing - SSL certificate error.

Thanks ufukcicekdev. I redeployed the all the modules and the role error disappears, it means that I can login into n8n now. However, there are still errors in Postgres and Redis as shown the my last reply.

Also, I would like to understand the cause of the crash and at least want a guarantee that the error has been fixed, so that it would not happen in the future, or else I will have to find alternative solutions for Railway.

Hi goonjanmall, thank you for sharing what you have done. I did not saw that threads and did not remove the said variable.

My workflow does not have an explicit webhook so I am not share the same experience with you. However, my telegram trigger is still working fine and I think it is a webhook too, if that is true, then I do not have that issue.

Yes, I'm having the same issues on the Redis side. But my flows are working, and I'm not having any problems.

Hey there, theres nothing to be worried about as far as those two unrelated issues, they just try connecting to the database to provide an internal management interface. It's been happening for a long time now and even I'm confused why, though I will reach out and see if I can get more information on this.

Thanks Jack for your input, much appreciated.

@ray-chenDo you have any comments? It has been a while.

9 days and still no reply.