I'm also getting very high latency on 30s to get a response back from env.VECTORIZE.upsert(vectors);

I'm also getting very high latency on 30s to get a response back from env.VECTORIZE.upsert(vectors); with a batch of 100 vectors of 768 size. I understood that this api was only queuing the work? and it's eventually consistent So why so high to call the add api?

Ccrunchy I'm also getting very high latency on 30s to get a response back from env.VECTOR...

crunchyOP•1/25/25, 11:59 PM

Ah I'm underestimating the JSON payload size of these vectors and the RPC overhead. I've if i limit the decimal places then I can get 20k rpm at batches of 100.

Rezigco•1/28/25, 1:14 PM

I've been reading and for my use case, I'm wondering if I'll need to combine both workers and the cloudflare api: I want each of my users to have their own index but I want the users of my users to be able to query my user's index and it could exceed the cloudflare api limit so I would like the queries/inserting to be done via cloudflare workers, but how can I dynamically pass in the name of the index

Utkarsh Saxena•1/30/25, 9:16 AM

Hi, I wanted to know about pricing of vectorization:
There are some confusing things in vectorization:

This is for a month:
Let's say I am making 10 queries of some documents where vector stored amount is: 10k and there are 100 those documents. and dimension is 376.
Then how will the queried vector dimension will be calculated:

if we calculate queried vector dimension of 10 queries for one document then: it will be: (10,000+ 10)376 = 3763760
Now for 100 documents it will be: 3763760100 = 376376000

but now it consider whole 100 documents queries at one level:
total queries will be: (100 ) (10) = 1000
Queries dimension for all 100 : (10,000+1000)*376 = 4136000

Which one is correct ?

garff•1/31/25, 2:59 PM

Hi, I have a worker deployed utilizing vectorize. It's deployed bound to a certain index but when I do my searches it also returns results from another index. Is this normal behaviour? Or what's up? Anyone had any troubles like that?

garff•1/31/25, 2:59 PM

Bindings in the settings looking under the dashboard says "Vectorize index
VECTORIZE
bge-large-en"

garff•1/31/25, 2:59 PM

And in swagger I get

garff•1/31/25, 2:59 PM

{
"success": true,
"embeddings": {
"count": 3,
"matches": [
{
"id": "15",
"score": 0.86592543,
"namespace": "findu-job-desc-test1",

garff•1/31/25, 3:00 PM

I feel certain that only the index you are currently utilizing should return results. Not other indexes. Or am I mistaken?

garff•1/31/25, 3:01 PM

And I'm trying to delete the index now - takes aaaages..

Mitya•1/31/25, 3:13 PM

Trying to get info on an index, as per the docs, via the following command:

npx wrangler vectorize info tmpdb

npx wrangler vectorize info tmpdb

But this gives:

Unknown arguments: info, tmpdb

crunchyOP•2/1/25, 10:36 PM

works for me

may be check your wrangler version and than the correct program is being triggered

crunchyOP•2/1/25, 10:38 PM

Is there any significance to the mutationIds? i.e. c0e522b1-3541-4aa8-a298-32c5c5fbf5b9. All i want to know is if there are in any kind of order. It's great seeing the latest one in status but which ones came inbetween? Similiar vector databases return a mutation number in a sequence]

RRafal is there anyone else running into `Error: VECTOR_QUERY_ERROR (code = 10000): Aut...

Bob•2/4/25, 4:21 PM

same for me too

RRafal is there anyone else running into `Error: VECTOR_QUERY_ERROR (code = 10000): Aut...

ggu•2/4/25, 4:41 PM

Hi Rafal, can you share your worker code with me so I can help debug it?

ggu•2/4/25, 4:41 PM

Also share the command you're using to launch it locally

RRafal Sure, here it is: https://gist.github.com/RafalWilinski/274d51bcd0b999eb161b00c3...

ggu•2/4/25, 4:59 PM

This code is running at the top-level, which seems strange to me. If you put the code within a fetch handler does the issue persist?

Thomas Ankcorn•2/4/25, 5:03 PM

Looks like something weird with getPlatformProxy. @Edmund Hung might know

Samko•2/4/25, 7:14 PM

I'm experiencing same issue with Error: VECTOR_GET_ERROR (code = 10000): Authentication errorError: VECTOR_GET_ERROR (code = 10000): Authentication error although it worked before. Already tried to upgrade to latest wrangler. I'm also getting:

✘ [ERROR] Error storing embeddings for undefined

  Error: VECTOR_UPSERT_ERROR (code = 999): vectorize.upstream_timeout
      at VectorizeIndexImpl._send (cloudflare-internal:vectorize-api:164:23)
      at async VectorizeIndexImpl.upsert (cloudflare-internal:vectorize-api:117:21)
      at async Object.queue
  (file:///Users/samko/Documents/Web/vecrio-api/workers/queue/src/index.ts:220:11) {
    [cause]: Error: vectorize.upstream_timeout
        at VectorizeIndexImpl._send (cloudflare-internal:vectorize-api:167:28)
        at async VectorizeIndexImpl.upsert (cloudflare-internal:vectorize-api:117:21)
        at async Object.queue

✘ [ERROR] Error storing embeddings for undefined

  Error: VECTOR_UPSERT_ERROR (code = 999): vectorize.upstream_timeout
      at VectorizeIndexImpl._send (cloudflare-internal:vectorize-api:164:23)
      at async VectorizeIndexImpl.upsert (cloudflare-internal:vectorize-api:117:21)
      at async Object.queue
  (file:///Users/samko/Documents/Web/vecrio-api/workers/queue/src/index.ts:220:11) {
    [cause]: Error: vectorize.upstream_timeout
        at VectorizeIndexImpl._send (cloudflare-internal:vectorize-api:167:28)
        at async VectorizeIndexImpl.upsert (cloudflare-internal:vectorize-api:117:21)
        at async Object.queue

Eric•2/6/25, 8:24 AM

https://developers.cloudflare.com/api/resources/vectorize/subresources/indexes/methods/list/ can anyone fix this link?

Eric•2/6/25, 8:25 AM

Perhaps this too? https://developers.cloudflare.com/api/resources/durable_objects/subresources/namespaces/methods/list/

Jerome•2/6/25, 9:20 AM

The service experienced an elevated rate of errors last 30 minutes, due to an R2 incident. It is now recovering and error rate is nominal. https://www.cloudflarestatus.com/incidents/pgz7g5xlpxzr

Cloudflare R2 Availability Issues

crunchyOP•2/6/25, 6:53 PM

Does vectorize have rate limits when called via workers? I'm unclear if the global limits only apply to the rest api? I'm luck getting those limits increased if i'm not an enterprise size customer. It's tricky to plan with the uncertainity

UUtkarsh Saxena Hi, I wanted to know about pricing of vectorization: There are some confusing th...

Utkarsh Saxena•2/7/25, 9:00 AM

Anyone knows about this?

dandoen•2/7/25, 11:43 AM

hi cloudflare team, are there any short term plans to support the multilingual BAAI/bge-m3 embedding model? the current BGE ones are english only. if it's not coming soon, i won't be able to use vectorize but if it's coming up then i can live with EN for now only!

Marco Aleixo•2/8/25, 9:35 PM

What is the roadmap for vectorize?
Do you plan to support multi dense vector like the ones from colpali?

RRafal is there anyone else running into `Error: VECTOR_QUERY_ERROR (code = 10000): Aut...

shoo•2/9/25, 5:30 PM

Facing the same issue - not sure what's causing it. query fails. Error: VECTOR_QUERY_ERROR (code = 10000): Authentication error

Bob•2/9/25, 7:27 PM

Hello, I have a question. It seems the indexing process takes few seconds. When we insert data, we get a mutation id. Is there any way to check against the mutation id if that batch got indexed or not?

beingmudit [WePlayDOS]•2/9/25, 8:42 PM

Damn! @Bob I joined this discord just to ask the same questions. Interesting that this is not just me experienceing indexing latency.

beingmudit [WePlayDOS]•2/9/25, 8:43 PM

Hey CF team!

We're building with Vectorize and noticed some interesting behavior around indexing latency and consistency. Here's what we're observing:

Indexing Latency:
Create index -> Success (immediate)
Add 2 vectors -> Success (immediate)
Search vectors -> No results for ~9 seconds
Search attempt timeline:
- 0s: No results
- 3s: No results
- 6s: No results
- 9s: Finally get results
Deletion Consistency:
After vectors are searchable
Delete vector "vec1" -> Success
Immediate search still returns both vectors including the deleted one
Subsequent searches also return the deleted vector

Test Case Details:

Index dimension: 32
Two simple test vectors:
- vec1: Array(32).fill(0.1)
- vec2: Array(32).fill(0.2)
Query vector: Array(32).fill(0.15)

Questions:

Is this expected latency for indexing?
Are there any best practices for handling the indexing delay?
Should deleted vectors remain searchable? If not, is there a known delay before deletions are reflected?
Are there any configuration options we should be using to optimize this?

Full test code available if needed. Thanks for any insights/help/

BBob Hello, I have a question. It seems the indexing process takes few seconds. When ...

Utkarsh Saxena•2/10/25, 5:23 AM

I think, You can check with getByIDs, as the insertion in index is async now in v2.

beingmudit [WePlayDOS]•2/10/25, 7:23 AM

Can you share the docs where it mentions something related to this? This seems like a big limitations given other services are super duper fast.

Utkarsh Saxena•2/10/25, 8:13 AM

Sure, You can see this: https://developers.cloudflare.com/vectorize/reference/transition-vectorize-legacy/#notable-changes
It is mentioned it takes few seconds to reflect and is asynchronous now.

Cloudflare Docs

Transition legacy Vectorize indexes · Vectorize

Legacy Vectorize (V1) indexes are on a deprecation path as of Aug 15, 2024. Your Vectorize index may be a legacy index if it fulfills any of the follwing crieria:

Sshoo Facing the same issue - not sure what's causing it. query fails. Error: VECTOR_Q...

cloudkite•2/10/25, 8:17 AM

Also getting this intermittently

beingmudit [WePlayDOS]•2/10/25, 9:19 PM

Thanks Utkarsh! This helps but i still can't beleive the 6-9 sec latency :/ Can anyone from Cloudflare team answer/confirm this?

Bbeingmudit [WePlayDOS]Thanks Utkarsh! This helps but i still can't beleive the 6-9 sec latency :/ Can ...

Jerome•2/10/25, 9:44 PM

Hi! yes the updates are processed asynchronously, and can take some time before being reflected in queries; 6 to 9s processing loop sounds like normal operation.

beingmudit [WePlayDOS]•2/10/25, 9:48 PM

Hey Jerome, thanks for getting back. Do you think if it makes sense to have it documented somewhere in the public facing docs? This async behaviour is really tricky to find.
Also, Do you know if other services do have similar latencies?

Jerome•2/10/25, 9:52 PM

. Do you think if it makes sense to have it documented somewhere in the public facing docs?

It's in the doc clearly stated for insert (same for upsert):
https://developers.cloudflare.com/vectorize/reference/client-api/

Jerome•2/10/25, 9:53 PM

Also, Do you know if other services do have similar latencies?

You mean, other cloudflare services?

beingmudit [WePlayDOS]•2/10/25, 10:33 PM

Ahhhh, I missed this page entirely. Thank you for your help.
No, I meant services like pinecone etc?

JJerome Hi! yes the updates are processed asynchronously, and can take some time before ...

Marco Aleixo•2/11/25, 2:35 AM

maybe we can have a wait=True for inserts?
I mean, I can implement the "confirm that the vector was indexed" pooling on my side, but maybe having this in the api itself is more straightforward...?!

UUtkarsh Saxena Hi, I wanted to know about pricing of vectorization: There are some confusing th...

Utkarsh Saxena•2/11/25, 5:18 AM

Can anyone help me with this

MMarco Aleixo maybe we can have a wait=True for inserts? I mean, I can implement the "confirm ...

garvitg•2/11/25, 5:09 PM

The Vectorize Index Info route provides information on what mutations Vectorize has completed processing: https://developers.cloudflare.com/api/resources/vectorize/subresources/indexes/methods/info/. It includes the mutation id and the timestamp of the last mutation block that Vectorize has processed.

Cloudflare API | Vectorize › Indexes › Get Vectorize Index Info

Interact with Cloudflare's products and services via the Cloudflare API

Sshoo Facing the same issue - not sure what's causing it. query fails. Error: VECTOR_Q...

cloudkite•2/11/25, 6:01 PM

Bump for visibility, also getting this. Hopefully it’s being looked into

Victor•2/14/25, 12:17 AM

https://discord.com/channels/595317990191398933/1040420029080018945/1319662221957271605

So I make an index.
I delete the index.
I make an index with same id.
verify new index info
I try to upsert vector - fails saying index is deleted
I retry the exact same upsert api request (no change) - success

beingmudit [WePlayDOS]•2/14/25, 6:18 AM

Yes because the operations are async. When you get the response. it is just an ack not the actual operation.

Victor•2/15/25, 1:49 AM

I've had it happen even as far as an insert 24h after making the index again. And I get index info after creating it and verify it was created correctly

deo time•2/22/25, 12:31 AM

Hello, I've had a site that uses vectorize for a little over a year now, however just now requests to it have started failing with (error) Error: VECTOR_QUERY_ERROR (code = 4009): Bad Request: the index is disabled(error) Error: VECTOR_QUERY_ERROR (code = 4009): Bad Request: the index is disabled. I've also checked and it looks like my index isn't there anymore, does anyone know why this is happening?

I'm also getting very high latency on 30s to get a response back from env.VECTORIZE.upsert(vectors);

Similar Threads

Similar Threads

Similar Threads