Hi, does anyone have a good guide to follow to dep...
# prisma1-community
a
Hi, does anyone have a good guide to follow to deploy a prisma 1 service to AWS? Is this the recommended one: https://v1.prisma.io/tutorials/deploy-prisma-to-aws-fargate-ct14#deploy-a-prisma-service-to-the-new-server
r
Hey @Abhischek 👋 Yes this is the recommended way to deploy to AWS Fargate 🙂
a
apologies I'm still new to this 🙂 So we have a container running in fargate already, I would like to push the latest changes to it. What's the best way to do that? I'm having trouble finding the ExternalURL from step 2.4. Any idea if this is no longer accurate?
r
For step 24, It will be available in the outputs section after you have deployed the Prisma Cluster stack.
a
Unfortunately not
r
Have you used this template?
a
I believe so, but let me try to create a new stack. This existing stack was created a while ago - not by me 🙂 So I just wanted to push the latest changes/docker file to it. but I will use the template to createa new stack - which will hopefully give me the URL I need to push changes to
r
Yeah it should. The URL is specified in the outputs in the one I have provided so it should be present if successfully created.
a
Hi @Ryan I managed to deploy to AWS ECS using the fargate yaml you pointed me to! Thx! We are running our own graphql server using
graphql-yoga
which I've packaged as another docker container. I need to supply the prisma endpoint URL to this container, they're part of the same AWS VPC. We are doing:
Copy code
"environment": [
        {
          "name": "ENDPOINT",
          "value": "<http://localhost:4466>"
        },
but that doesn't seem to work. Do you know how to address the prisma container in the
prisma.yml
when they're deployed as a separate container?
r
If I have understood correctly, the Prisma server and your application (GraphQL Yoga) are on separate containers right? So shouldn't the endpoint URL be created Prisma endpoint that you get after deploying the stack instead of
localhost
?
a
you're right - the issue wasn't that I couldn't connect to prisma, but that my prisma container wasn't booting because it can't connect to mongo. have you tried this before? I have VPC peering setup and can see that prisma resolves my mongo URI to a local dns:
Copy code
INFO] Fatal error during deployment worker initialization: com.mongodb.MongoTimeoutException: Timed out after 30000 ms
r
Unfortunately I haven't deployed Prisma in a VPC peering setup so it would be difficult for me to debug the issue.
🙏🏾 1
a
thank you you've been helpful already 🙂
Maybe this will help someone else from going bald 🙂 If you are using networking mode
awsvpc
then you can access other containers you are deploying by the
localhost
interface. So I passed http://localhost:4466 to my graphql - server to connect to prisma. More details here: https://docs.aws.amazon.com/AmazonECS/latest/developerguide/task-networking.html
🙌 1
o
I personally containerized and deployed on Elastic Beanstalk. Really simple to manage and scale
a
Hi @Omar Nasr would you be able to share your mongo configuration that you pass into
PRISMA_CONFIG
? Assuming that you are using Mongo