After upgrading mysql to 8.0.30, prisma is no longer able to connect:

Error querying the database: Server error: `ERROR 08S01 (1043): Bad handshake'

I am unable to write a bug report right now, so just dropping a quick warning here, before you upgrade your mysql server.

TL;DR: If all calls to the prisma client are synchronous, are queries sent to a single instance of the Prisma client guaranteed to be in the same connection/session when using postgres? Background: I’d like to set a transactional advisory lock where i need some processing between queries. • I can’t use the Sequential Operations feature because i need processing between queries. • I can’t use the Interactive transactions feature because this is for production. • I’d prefer to use a transactional and not a general advisory lock on the db, because then we need to worry about cleaning up orphaned advisory locks in the unlikely but real scenario of a node dying without ‘unlocking’. Why can’t I just use this? Are queries sent to a single instance of the Prisma client not guaranteed to be in the same connection? Running in my repl, it appears to be the same connection.

prisma.$queryRaw`BEGIN`
prisma.$queryRaw`pg_advisory_xact_lock(5)`
// code, processing
prisma.$queryRaw`pg_advisory_xact_unlock(5)`
prisma.$queryRaw`COMMIT`

Hi everyone. I’m stuck with a slow query I’d like to speed up. TLDR; this query is super slow. What should I do to speed this query up? Can I do it by adding an index?

prisma.concept.findMany({
    where: {
      dictionaryId: Number(dictId),
      name: {contains: foo}
    },
    include: {
      _count: {
        select: {
          declarations: true, // including this in the query makes it super slow
        },
      },
    }
  })

I have a concept table and a declaration table. Concept and declaration have a many to many relationship. The declaration table has more than 500k rows, same goes for concept table. longer explanation: https://www.loom.com/share/ff8d3034898e4ad08b8d19f044d500fe

Hi I’m currently trying to upgrade to Prisma 4 and I’m having a little casting issues in some

$queryRaw()

Example query:

const rows = await this.prismaClient.$queryRaw<
  { id: string; location: string }[]
>(Prisma.sql`
  SELECT id, ST_AsGeoJSON(location) as location,
  FROM something
  WHERE id IN (${Prisma.join(ids)})
`);

I get some type errors I didn’t have in previous versions:

ERROR: operator does not exist: uuid = text

I’ve tried various way to cast my list of

ids

to uuids, but to no avail so far.

IN (${Prisma.join(ids)}::uuid[]) IN (${Prisma.join(ids)})::uuid[]

I wouldn’t mind some help there since I might be missing something simple 😅

Hey! I’m struggling with a strange error creating a new item in a collection that I can’t seem to find anything on. Best explained with my schema files, the failing create call, and the errors that ensue. Here’s my schema for the idea of a ‘report’, which has to belong to a ‘team’.

model Report {
  id            String        @id @default(auto()) @map("_id") @db.ObjectId
  ...
  team          Team          @relation(fields: [teamId], references: [id])
  teamId        String 
  ...
}

Here’s the Team:

model Team {
  id                   String             @id @map("_id")
  ...
  reports              Report[]
  ...
}

I’ve tried creating a number of different ways. I thought the right way was as such:

const report = await prisma.report.create({
        data: {
            id : newId.toString(),
            teamId:        teamId,
            ...
        },
    });

Which gives error:

"Parse errors: [Query parsing/validation error at Mutation.createOneReport.data.ReportCreateInput.team: A value is required but not set"

Which I interpret to mean it’s missing the ‘team’ object. So I tried:

const report = await prisma.report.create({
        data: {
            id : newId.toString(),
            ...
            teamId:        teamId,
            team: {
                connect: {
                    id: teamId,
                },
            },
            ...
        },
    });

Which errors out telling me I shouldn’t be including team Id. (will include error in thread). But if I don’t invlude teamId it errors out telling me I should be using teamId. Very confusing since I used to be able to just add teamId in these types of situations

Hello 👋 Has anyone managed to build some kind of query builder or query composer? I've found this discussion in Prisma's repo, but it doesn't really have any answer: https://github.com/prisma/prisma/discussions/12047#discussioncomment-3501408 I've tried to describe my use-case in the comment there. Basically I'd like several small repository/service methods that take care of a single responsibility, and can be combined together into a single strongly-typed query/result. I'd be grateful for really any resource on how to design a DRY, reusable repositories/services with Prisma. Thanks!

Hey Gautam! What versions of Prisma and TypeScript are you using?

hey folks, I'm trying something new ... MongoDB i've managed to convert my schema into a mongo "happy" format, but when I try to write data, I'm getting some errors when I'm linking to documents in other collections

//schema

model activityLog {
  activityLogId      String                @id @map("_id") @default(uuid())
  activityLogType    activityLogType       @relation(fields: [activityLogTypeId], references: [activityLogType])
  activityLogTypeId  String                @db.ObjectId()
  activityLogMessage String                @db.String
  activityCompleted  Boolean               @default(false)
  createdAt          DateTime              @default(now()) @db.Timestamp
  updatedAt          DateTime              @default(now()) @db.Timestamp
}

model activityLogType {
  activityLogType     String                @id @map("_id")  @db.String
  createdAt           DateTime              @default(now()) @db.Timestamp
  updatedAt           DateTime              @default(now()) @db.Timestamp
  activityLog         activityLog[]
}

when I try to do

await prisma.activityLog.create({
    data: {
      activityLogType: {
        activityLogType: "Start"
      },
      activityLogMessage: "Start Seeding",
      activityCompleted: true
    }
  })

The

Start

document already exists in the

activityLogTypes

collection it complains that the

activityLogType

does not exist schema I'm probably doing something wrong, and its likely a n00b error for Mongo, but any help would be suggested.

Hi there - I noticed the release today with preview support for field reference support on query filters. Curious if this is a precursor or helps in any way to address this issue (sorting by foreign key fields): https://github.com/prisma/prisma/issues/5837

Looks like it’s working as intended. The only way to make this work is to select the value on both, otherwise there’s only one case where birthdate is a value so Typescript can’t safely let you use the property.

Hi friends, i wanted to ask about naming convention of models and how it relates to reusing code and pulling/introspection. Let me explain the scenario. Folder structure

/core/prisma/schema.prisma
/lambda/prisma/schema.prisma

Core - This folder contains some core logic and is often "shared" code used by folder

lambda

. Lambda folder is a serverless function more on that later.

core/prisma/schema.prisma

model Collection {
    id                String    @id @default(auto()) @map("_id") @db.ObjectId
    network           String
    contract_address  String    @unique
    @@map("collections")
}

The convention for declaring a model here is familiar to the docs. https://www.prisma.io/docs/concepts/components/prisma-schema - Singular with a capital first letter.

core/models/collections.ts

Inside of this file im doing some common functionality like finding, inserting etc. Note: the model name that is being used "collection". Generated by the prisma library based on schema.prisma.

export const FindCollections = async () => {
    try {
        const contracts = await PrismaService.collection.findMany();

        return [contracts, null] as const;
    } catch (err) {
        Logger.error({ kind: "FindCollections", error: err.name, stack_trace: err.stack });
        return [null, err] as const;
    }
};

Up until now, there are no problems. Now lets introduce my lambda. I have this lambda folder which has my serverless function im going to deploy and i want to use "FindCollections" function within the core folder which has already been created. However in my lambda folder, i need to introspect the database and package the engine. but when i introspect the database the model name becomes a problem.

lambda/prisma/prisma.schema

model collections {
  id                String   @id @default(auto()) @map("_id") @db.ObjectId
  contract_address  String
  ...
}

You can now see its all lowercase with plural. Now with this introspected schema, my existing "FindCollections" function no longer works, because the model name is different.

InsertCollections","stack_trace":"Error: Could not find mapping for model Collection\n

What is the suggestion for this usecase? There are 2 solutions i can think of but im wondering what you guys think.

Hi! Is there anyway to use the prisma CLI to check if the remote schema matches the local one?

Looks like someone accidentially packaged prisma with the debian libquery_engine in the

internals

dist

folder 😕

Looks like you are hitting this bug: https://github.com/prisma/prisma/issues/5715 Please add +1 in the issue to get it up the priority list.

FYI opened a ticket for the accidentially bundled libquery_engine: https://github.com/prisma/prisma/issues/15244

Hi friends, i am getting the following error when trying to insert into mongodb: unfortunately prisma does not indicate the name of field that is the problem in this error message.

16 try {
  17     const response = await PrismaService.$transaction(
  18         payload.map((p) =>
→ 19             PrismaService.salesByContractAddress.upsert(
Error parsing GraphQL query: query parse error: Parse error at 23:14
number too large to fit in target type

is there some way i can determine which field in the schema is the problem?

Does prisma support mongodb's

Decimal128

?

Hi! new to prisma and trying to understand some basics, trying to do something simple (or seems should be simple) but running into some typing crazyness.

async update(updatedBy:Principal, data: Prisma.BlogUpdateInput): Promise<BlogEntry> {
        data.updatedBy = updateBy.principalId;
        return await this.prisma.blogEntry.update({
                                where: {
                                    id: data.id
                                },
                                data: data
                            });
    }

but this yields me an syntax error stating:

Type 'string | StringFieldUpdateOperationsInput' is not assignable to type 'string'.
  Type 'StringFieldUpdateOperationsInput' is not assignable to type 'string'.ts(2322)
index.d.ts(4311, 5): The expected type comes from property 'id' which is declared here on type 'BlogEntryWhereUniqueInput'

BlogUpdateInput.id is just a string... what "generated" type should I be using to on my

update()

method that wraps the call to prisma? This is a bit confusing

Hi, is it good practice to check-in the auto-generated clients generated by

npx prisma generate

to Git, or should I run this command as part of my

prestart

step in

npm

? Context: I am deploying my web app in an air-gapped system and the

npx prisma generate

tried to fetch some files from an online URL and failed miserably. Now I’m debating between: 1. Runing

npx prisma generate

as part of the Docker build - to online fetch the auto-generated files 2. Checking-in the prisma client to Git

In simpler words, should Prisma client be version controlled (Git) or auto-generated through

npx prisma generate

on

npm run dev

?

I now realize that what failed is the query engine download, I could cache it locally or simply run

npx prisma generate

as part of the Docker build step and have it cached as a side effect.

Hey 🙌 Does someone know how to check connection status with prisma client (like an healthcheck) ?

Hi there! New to this channel so please redirect me if this should go elsewhere. I'm using prisma with a mysql planetscale backend. I'm getting an error about ROW_NUMBER() in a

prisma.$queryRaw

Something like this, which works fine on my planetscale console:

const query = `select ROW_NUMBER() over() FROM contacts`;

Results in the error:

Code: `1105`. Message: `unknown error: syntax error at position 26`

Anyone know why prisma can't seem to handle

ROW_NUMBER()

function?

Hi! Can I perform a full text search with cockroach db?

If I upgrade from prisma 3.x to 4.x will it stop performing wasted queries to get the results of inserts?

HI there! Currently debugging some crazy stuff with generating a prisma client with a custom output path into a monorepo man that's a lot of buzzwords I have found several open issues (and some old closed ones) with people who seem to have the same problem, wondering if this bug isn't a priority and I should just switch everything off of prisma? Issues are here: Technically Closed: https://github.com/prisma/prisma/issues/10512, Open: https://github.com/prisma/prisma/issues/13233, https://github.com/prisma/prisma/issues/12588, https://github.com/prisma/prisma/issues/10433 (my exact issue) I have figured out how to fix this sort of! I just had to manually edit the generated index.js file 😂 Details in the thread Hoping to get a real fix for this ❤️

How do I explicitly get a new connection from a prisma object?

is there a way to dynamically create or notify the prisma client to new tables without having to rebuild the entire client? the example specifically that comes into play is if i am creating more or less a CMS that allows the individual to create a data model (think more amongst the lines of strapi / contentful), i would like to generate out a table structure for that and then allow prisma to understand it without having to update the schema file and then run migrations and generate the client once again, i am sure some form of generation needs to happen but to be able to dynamically add in additional tables without the full generator process running again to create the underlying client would be nice.

Hello - can someone please take a look this thread? I believe i may have posted it in the wrong place https://prisma.slack.com/archives/CA491RJH0/p1663781800291489

👋 Hello everyone! To streamline support for our Community, all questions about the Prisma ORM (Client, Studio, Migrate) will now be answered in #orm-help channel, and this channel will be archived next week is now archived. We hope that this will allow us to help you more quickly and share the knowledge among all our community members. 😊 If you have any pending questions / issues, feel free to resolve them in the current threads or move the conversation into #orm-help! 🧑‍💻 Happy coding!