Prisma•8mo ago

is there still no option to manually invalidate the cache?

with accelerate, is this even in development?

8 Replies

Hey @sylvester stallone 👋🏾 , To manually invalidate the cache you have to go to the platform console. You can also create add a cacheKey field in your model such as,

model User {
  id @id @unique 
  name String
  description String
  cacheKey  String @unique
}

model User {
  id @id @unique 
  name String
  description String
  cacheKey  String @unique
}

You would have to store the cacheKey in some key value store and update it on every invalidation.

// change cache key value
const updateCacheKey = await updateCacheKey(userId)

await prisma.user.update({
  data: {
     updateCacheKey,
     ...otherUpdates
  }
})

// change cache key value
const updateCacheKey = await updateCacheKey(userId)

await prisma.user.update({
  data: {
     updateCacheKey,
     ...otherUpdates
  }
})

During reads,

const updateCacheKey = await getCacheKey(userId)
await prisma.user.findUnique({
  data: {
    where: {
      id: userId,
      cacheKey: cacheKey
    }
  },
  cacheStrategy: {
     ttl: 100
  }
})

const updateCacheKey = await getCacheKey(userId)
await prisma.user.findUnique({
  data: {
    where: {
      id: userId,
      cacheKey: cacheKey
    }
  },
  cacheStrategy: {
     ttl: 100
  }
})

This would guarantee invalidations on every update, even on development. Let me know if this helps!

sylvester stalloneOP•8mo ago

how would this work on a findMany @Ankur Datta also when is cache invalidation shipping in accelerate? is there an RFP we could trakc?

Ankur Datta•8mo ago

how would this work on a findMany

It would be a bit tricky with findMany, but it's doable.

await prisma.user.findMany({
  data: {
    where: {
      id: {
         in: [1, 2, 3]     
      },
      OR: {
         cacheKey: cacheKey // No model would have this cache key, but you can update it regardless, so that Accelerate views it as a different query
// The OR would ensure that the correct data is still returned
      }
    }
  },
  cacheStrategy: {
     ttl: 100
  }
})

await prisma.user.findMany({
  data: {
    where: {
      id: {
         in: [1, 2, 3]     
      },
      OR: {
         cacheKey: cacheKey // No model would have this cache key, but you can update it regardless, so that Accelerate views it as a different query
// The OR would ensure that the correct data is still returned
      }
    }
  },
  cacheStrategy: {
     ttl: 100
  }
})

Let me know if I need to elaborate 😄 .

Also when is cache invalidation shipping in accelerate? is there an RFP we could trakc?

We have no timelines for this as of yet. I'll let you know once we have something 🙏🏾 .

colhoop•4mo ago

The cache key method is interesting and probably something I will have to do on some of my queries for now. But I really think a built-in method for invalidating query cache should be near the top of the roadmap's priority list. I would consider a feature like this MVP necessary Also an option to refresh the cache in the background on demand please 🙂

Ankur Datta•4mo ago

@colhoop we're working on cache invalidation for Accelerate 😄 . Keep an eye on the #announcements channel for any news on this 👀 .

Ved•3mo ago

Hi, any update on this?

Raj•3mo ago

Been waiting on on-demand invalidation since Accelerate was first announced.

Ankur Datta•3mo ago

Hey @Ved and @Raj 👋🏾 , thanks for letting your interest known. We'll reach out to you once the feature is out 😄 .

Gaming

Programming

is there still no option to manually invalidate the cache?