Fix Rallly start.sh

[andreasdueren]

@girish said in Fix Rallly start.sh:

Strange, in bash, both are the same. Given that CLOUDRON_MAIL_SMTP_PORT is just a number.

Yeah no idea. But these were my logs (I think, pulled this out of my clipboard manager):

Oct 18 08:14:43 => Healthcheck error: Error: connect ECONNREFUSED 172.18.16.158:3000
Oct 18 08:14:43 => Healthcheck error: Error: connect ECONNREFUSED 172.18.16.158:3000
Oct 18 08:14:43 => Healthcheck error: Error: connect ECONNREFUSED 172.18.16.158:3000
Oct 18 08:14:43 => Healthcheck error: Error: connect ECONNREFUSED 172.18.16.158:3000
Oct 18 08:14:44 /app/pkg/start.sh: line 54: prisma: command not found
Oct 18 08:14:44 => Updating permissions
Oct 18 08:14:44 => prisma migrate
Oct 18 08:14:55 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:15:03 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:15:13 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:15:23 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:15:33 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:15:44 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:15:48 /app/pkg/start.sh: line 54: prisma: command not found
Oct 18 08:15:48 => Updating permissions
Oct 18 08:15:48 => prisma migrate
Oct 18 08:15:57 => Healthcheck error: AbortError: The operation was aborted
Oct 18 08:16:03 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:16:13 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:16:23 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:16:33 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:16:43 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:16:49 => Updating permissions
Oct 18 08:16:49 => prisma migrate
Oct 18 08:16:49 /app/pkg/start.sh: line 54: prisma: command not found
Oct 18 08:17:03 => Healthcheck error: AbortError: The operation was aborted
Oct 18 08:17:07 => Healthcheck error: AbortError: The operation was aborted
Oct 18 08:17:13 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000
Oct 18 08:17:23 => Healthcheck error: Error: connect EHOSTUNREACH 172.18.16.158:3000

[james]

Hello @andreasdueren
The error is:

Oct 18 08:16:49 /app/pkg/start.sh: line 54: prisma: command not found

[nebulon]

That error is even more peculiar. So the start.sh just runs prisma which should be in the path and part of the app image: https://git.cloudron.io/packages/rallly-app/-/blob/main/start.sh?ref_type=heads#L54

Are there any variables in the /app/data/env which may interfere with prisma or the PATH ?

[andreasdueren]

@nebulon said in Fix Rallly start.sh:

That error is even more peculiar. So the start.sh just runs prisma which should be in the path and part of the app image: https://git.cloudron.io/packages/rallly-app/-/blob/main/start.sh?ref_type=heads#L54

Are there any variables in the /app/data/env which may interfere with prisma or the PATH ?

I haven't touched the devault ENV

SECRET_PASSWORD='STRING'
NEXTAUTH_SECRET='STRING'

# Comma separated list of email addresses that are allowed to register and login.
# You can use wildcard syntax to match a range of email addresses.
# Example: "john@example.com,jane@example.com" or "*@example.com"
ALLOWED_EMAILS=

[keldorn]

Hello, same error here.

I did reinstall the application once, from scratch. At first, it did work, until it did not.

@andreasdueren suggested fix worked for me. Thank you.

[andreasdueren]

@keldorn yes that’s what I did too. Reinstalled them broke again

[nebulon]

would be great to have a way to reproduce this. Especially since prisma is just part of the image itself.

Can you for example run which prisma? It should print /tmp/pnpm/prisma if the PATH env is correct. Though I cannot quite see why that wouldn't be the case unless the image is somehow broken locally. Like https://git.cloudron.io/packages/rallly-app/-/blob/main/start.sh?ref_type=heads this is all the app does when the container starts.

[andreasdueren]

@nebulon im Sorry I don't know what to tell you except for this is what fixed it. which prisma returns nothing:

root@37e3d748-cecb-49e4-b218-88d23a2b37a4:/app/pkg# which prisma
root@37e3d748-cecb-49e4-b218-88d23a2b37a4:/app/pkg# 

[keldorn]

Some logs as well below.

So until ~19:34 this is the original configuration.

Then I did reboot to emergency mode, fixed the brackets, and then rebooted.

When I reboot the container and it works, which prisma correctly prints the path. If the host is rebooted, then it fails again (as the fix / workaround is, as expected, not retained)

I assume that adding the brackets would have no adverse effects.

Oct 27 19:34:15 /app/pkg/start.sh: line 54: prisma: command not found
Oct 27 19:34:15 => Setting up OIDC
Oct 27 19:34:15 => Updating permissions
Oct 27 19:34:15 => prisma migrate
Oct 27 19:35:12 Repair mode. Use the webterminal or cloudron exec to repair. Sleeping
Oct 27 19:37:55 => Setting up OIDC
Oct 27 19:37:55 => Updating permissions
Oct 27 19:37:55 => prisma migrate
Oct 27 19:37:58 2025-10-27T19:37:58+01:00
Oct 27 19:37:58 2025-10-27T19:37:58+01:00
Oct 27 19:37:58 2025-10-27T19:37:58+01:00
Oct 27 19:37:58 > Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 0%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 1%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 3%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 4%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 6%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 7%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 8%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 11%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 16%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 20%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 22%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 25%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 27%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 28%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 31%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 33%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 35%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 37%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 39%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 43%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 46%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 48%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 50%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 55%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 60%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 64%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 69%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 73%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 78%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 82%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 88%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 91%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 93%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 95%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 97%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 98%> Downloading Prisma engines for Node-API for debian-openssl-3.0.x [] 100%Environment variables loaded from .env
Oct 27 19:37:58 Datasource "db": PostgreSQL database "dbe49f54e745c046d38f7e4a0e61387c84", schema "public" at "postgresql:5432"
Oct 27 19:37:58 Prisma schema loaded from ../../../pkg/prisma/schema.prisma
Oct 27 19:37:58 No pending migrations to apply.
Oct 27 19:37:58 96 migrations found in prisma/migrations
Oct 27 19:37:58 => Starting Rallly
Oct 27 19:37:59 2025-10-27T19:37:59+01:00
Oct 27 19:37:59 ▲ Next.js 15.4.6
Oct 27 19:37:59 ✓ Ready in 430ms
Oct 27 19:37:59 ✓ Starting...
Oct 27 19:37:59 - Local: http://e49f54e7-45c0-46d3-8f7e-4a0e61387c84:3000
Oct 27 19:37:59 - Network: http://e49f54e7-45c0-46d3-8f7e-4a0e61387c84:3000

[nebulon]

Still wasn't able to reproduce to debug this. However, looking at it just now with fresh eyes on the topic, I think the error comes from the fact that pnpm for some reason is setup to use /tmp which I didn't realize before:

ENV PNPM_HOME="/tmp/pnpm"
ENV PATH="$PNPM_HOME:$PATH"

Not yet sure why this works in the first place, since /tmp is mounted as a temporary volume into the container as well as regularily purged, since it is a temporary storage. Hopefully better understand the root issue tomorrow and the provide a fix for it.