Connections from repl.it repls randomly failing with various errors

Bug description

Prisma connections from a lot of my repl.it repls randomly fail with various errors due to unknown reasons.

Here are the errors I’ve seen (Descending order of frequency):

[ERROR] : 
Invalid `prisma.workspace.findUnique()` invocation:
  Can't reach database server at `ec2-50-19-32-202.compute-1.amazonaws.com`:`5432`
Please make sure your database server is running at `ec2-50-19-32-202.compute-1.amazonaws.com`:`5432`.

Invalid `prisma.workspace.findUnique()` invocation:
  Error creating a database connection. (Operation timed out (mobc timeout))
    at new AuthorizationError (/home/runner/InfoBot-old-2/node_modules/@slack/oauth/dist/errors.js:71:28)
    at InstallProvider.<anonymous> (/home/runner/InfoBot-old-2/node_modules/@slack/oauth/dist/index.js:158:31)
    at step (/home/runner/InfoBot-old-2/node_modules/@slack/oauth/dist/index.js:44:23)
    at Object.throw (/home/runner/InfoBot-old-2/node_modules/@slack/oauth/dist/index.js:25:53)
    at rejected (/home/runner/InfoBot-old-2/node_modules/@slack/oauth/dist/index.js:17:65)

Invalid `prisma.channel.upsert()` invocation:
  Timed out fetching a new connection from the pool. Please consider reducing the number of requests or increasing the `connection_limit` parameter (https://www.prisma.io/docs/concepts/components/prisma-client/connection-management#connection-pool). Current limit: 5, amount of connections currently in use: 0.
    at Object.asCodedError (/home/runner/HackClubKarma/node_modules/@slack/bolt/dist/errors.js:23:12)
    at App.handleError (/home/runner/HackClubKarma/node_modules/@slack/bolt/dist/App.js:442:43)
    at App.processEvent (/home/runner/HackClubKarma/node_modules/@slack/bolt/dist/App.js:435:25)
    at processTicksAndRejections (internal/process/task_queues.js:97:5)

Note

I’ve already spoken to folks at repl.it and @nikolasburk – here’s some context:

How to reproduce

Not always reproducible. Occurs randomly and can only fixed by creating a fresh copy of (forking) the repl

Affected repls:

• https://repl.it/@KhushrajRathod/InfoBot-old-3
• https://repl.it/@KhushrajRathod/InfoBot-old-2
• https://repl.it/@KhushrajRathod/InfoBot-old
• https://repl.it/@KhushrajRathod/LightningDelete-old
• https://repl.it/@KhushrajRathod/The-Overseer-Frontend

Expected behavior

• Prisma can connect to the database (since it works locally and via psql)

Prisma information

datasource db {
    url      = env("DATABASE_URL")
    provider = "postgresql"
}

generator client {
    provider = "prisma-client-js"
}

model Workspace {
    id           String @id
    installation String
}

Environment & setup

• OS: Linux
• Database: PostgreSQL
• Node.js version: v12.16.1
• Prisma version: 2.14.0

[khrj]

If you have the time, it would be very helpful if you could try to explicitly call the disconnect in your code. This way we can ensure there are no dangling connections and see if the problem persists. A pattern for doing that is detailed here: https://www.prisma.io/docs/concepts/components/prisma-client/connection-management#calling-disconnect-explicitly

Hi, I’ll surely try that but I’m not sure how long it will take to know if the problem is solved, since the issue randomly reproduces

[garrensmith]

Hey @KhushrajRathod, I think this could be fixed. Would you be able to try with out with prisma@dev