I have a docker container that contains 2 images. My app and my database (postgres) They are both running, the app on :3000 and postgres on :5432 . When i run npx prisma migrate dev --name init
I get the following error :
Environment variables loaded from .env
Prisma schema loaded from lib/database/myschema.prisma
Datasource "db": PostgreSQL database "sales", schema "public" at "postgres:5432"
Error: P1001: Can't reach database server at `postgres`:`5432`
Please make sure your database server is running at `postgres`:`5432`.
If I check, the database is created, but no tables are created inside of it. I don't understand the error, because if the database is created obviously the path is correct and prisma is able to at least run a querry.
I am also connected to the database with PgAdmin and I'm able to run querry and create tables no problem.
Also, if I stop the app image but leave the postgres image up and change the database url to postgresql://admin:admin@localhost:5432/sales
I can run the migrations from my local repo directly with no issues and the tables are created.
Here is my schema.prisma
datasource db {
url = env("DATABASE_URL")
provider = "postgresql"
}
generator client {
provider = "prisma-client-js"
binaryTargets = ["native", "linux-arm64-openssl-1.1.x"]
}
model Sale {
id Int @id @default(autoincrement())
saleId Int
uniqueSaleId String
createdAt DateTime
completedAt DateTime
storeId Int
vendorId Int
customerId Int
currency String
payment String
total Float
productSolds ProductSolds[]
accountId String
account Account @relation(fields: [accountId], references: [id], onDelete: Cascade)
@@unique([saleId, accountId])
}
model ProductSolds {
id Int @id @default(autoincrement())
productId Int
productSize Int
quantity Int
productPrice Float
productCurrency String
vat Float
package Int
points Int
discount Float
stockWithdrawal Int
creditNote Int
creditNoteId Int
productComments Float
serialNumber Float
lineItemIdReturn Int
dateReturn Int
kitchenPos Int
productSupplyPrice Float
lineItemId Int
detailCommandeId Int
saleId Int
sale Sale @relation(fields: [saleId], references: [id], onDelete: Cascade)
}
model EndpointCalled{
id Int @id @default(autoincrement())
accountId String
year Int
month Int
day Int
storeId Int
@@unique([accountId,year,month,day,storeId])
}
model Account {
id String @id
Sale Sale[]
}
my DATABASE_URL in my .env is postgresql://admin:admin@postgres:5432/sales
although I change it to localhost for testing purposes when I'm running the app outside of docker (and it works) .
I though the issue might be with the service name inside docker, so I specified it with container_name: postgres
My docker-compose :
version: "3.7"
services:
node:
build:
context: .
dockerfile: devops/docker/dockerfiles/Dockerfile
target: development
volumes:
- .:/home
- ${GCP_SA_KEYS_PATH}:/gcp-sa-keys
command: shell
environment:
adminBearerTokens: 1234
GOOGLE_APPLICATION_CREDENTIALS: /gcp-sa-keys/${GCP_SA_KEY}
DATABASE_URL: ${DATABASE_URL:-postgresql://admin:admin@postgres:5432/sales}
HIBOUTIK_USER: ${HIBOUTIK_USER}
HIBOUTIK_API_KEY: ${HIBOUTIK_API_KEY}
depends_on:
- postgres
ports:
- "3000:3000"
postgres:
image: postgres:14.1-alpine
container_name: postgres
environment:
POSTGRES_USER: admin
POSTGRES_PASSWORD: admin
ports:
- "5432:5432"
In case this is relevant, I am running the containers on a macbook air M1 and had to modify the image a bit to make it run.
CodePudding user response:
The database url is not correct. You need something like this:
DATABASE_URL=driver://username:password@hostname:port/database_name
And you have
DATABASE_URL: ${DATABASE_URL:postgresql://admin:admin@postgres:5432/sales}
Instead of the hostname, you use "postgres". Unless you have a host like that, it's not going to work. And the error message tells us that you don't have this host. "localhost" works when connecting local, otherwise your need the ip address or hostname. That depends on your setup.
CodePudding user response:
I found the solution here
The fix is to add ?connect_timeout=300
at the end of the string. So in my case postgresql://admin:admin@postgres:5432/sales?connect_timeout=300
It seems to be an issue with node 16, prisma, and mac M1. why does adding connect_timeout=300
work ? I have no idea and I couldn't find an answer online. But I could run my migrations and everything now runs smoothly.