SQLite times out during query execution when using Promise.all() / concurrent

Bug description

I’m attempting to await multiple queries at once using Promise.all(), something which works fine on MySQL, however on SQLite this gives the following error

touch(): Error: 
Invalid `prisma.session.update()` invocation:


  Error occurred during query execution:
ConnectorError(ConnectorError { user_facing_error: None, kind: ConnectionError(Timed out during query execution.) })
Error: 
Invalid `prisma.session.update()` invocation:


  Error occurred during query execution:
ConnectorError(ConnectorError { user_facing_error: None, kind: ConnectionError(Timed out during query execution.) })
    at cb (xxx\node_modules\@prisma\client\runtime\index.js:36956:17)

I’m hoping to use Promise.all() to execute and await multiple queries at once as performance optimization. I have also found the library prisma-session-store to use a similar pattern thus also resulting in this error.

It is also worth mentioning the error isn’t always thrown. It seems fairly inconsistent when the error is and isn’t thrown, however most of the time it is.

How to reproduce

1. Pass multiple concurrent queries into Promise.all()
2. Error is usually thrown

Expected behavior

The queries complete concurrently and the Promise.all() resolves. This is the behavior in MySQL.

Prisma information

Example:

await Promise.all([
	database.tableOne.update({ ...your query... }),
	database.tableTwo.update({ ...your query... })
]);

Environment & setup

• OS: Windows
• Database: SQLite
• Node.js version: v14.17.1

Prisma Version

prisma                : 2.30.3
@prisma/client        : 3.1.1
Current platform      : windows
Query Engine (Binary) : query-engine b8c35d44de987a9691890b3ddf3e2e7effb9bf20 (at node_modules\@quixo3\prisma-session-store\node_modules\@prisma\engines\query-engine-windows.exe)
Migration Engine      : migration-engine-cli b8c35d44de987a9691890b3ddf3e2e7effb9bf20 (at node_modules\@quixo3\prisma-session-store\node_modules\@prisma\engines\migration-engine-windows.exe)
Introspection Engine  : introspection-core b8c35d44de987a9691890b3ddf3e2e7effb9bf20 (at node_modules\@quixo3\prisma-session-store\node_modules\@prisma\engines\introspection-engine-windows.exe)
Format Binary         : prisma-fmt b8c35d44de987a9691890b3ddf3e2e7effb9bf20 (at node_modules\@quixo3\prisma-session-store\node_modules\@prisma\engines\prisma-fmt-windows.exe)
Default Engines Hash  : b8c35d44de987a9691890b3ddf3e2e7effb9bf20
Studio                : 0.423.0

[HartS]

I’m seeing the same issue using prisma with sqlite, unrelated to seeding or Promise.all();

When executing 5 updates in parallel (not even on the same records, just the same table), it seems to occur with just one of the records. In my case I’m running the updates in parallel by using forEach over a set of records I need to update, but I’m going to work around it by using a for loop with an await inside.

[noxware]

I’m not using Promise.all but I have a CLI app that runs n worker functions concurrently and I’ve been fighting this problem all night.

I’ve written a function serializer and wrapped all prisma calls with it and it worked. But then I read the ?connection_limit=1 trick which worked without my serialization stuff, which is better.

I’ve started using Prisma a month ago and I feel like SQLite support is in general very poor. It’s very demotivating.

Extra note: Ensure that you don’t have zombie node.js process in background accessing to the database or the connection_limit trick will not work 😅 I’m saying this because after hours of trying every possible fix and configuration, the final problem was that I accidently had 3 zombie nodejs processes in the background touching the db.