flowiseai / flowisedocs Goto Github PK

From the documentation (vector-stores/supabase.md), running the script is coming back with an error "Invalid SQL query". This seems to be from the function section as the table script executes fine. I am not good enough with SQL to understand what's not working well, but wanted to raise it to review.

Here is the script directly from your documentation:

-- Enable the pgvector extension to work with embedding vectors
create extension vector;

-- Create a table to store your documents
create table documents (
  id bigserial primary key,
  content text, -- corresponds to Document.pageContent
  metadata jsonb, -- corresponds to Document.metadata
  embedding vector(1536) -- 1536 works for OpenAI embeddings, change if needed
);

-- Create a function to search for documents
create function match_documents (
  query_embedding vector(1536),
  match_count int DEFAULT null,
  filter jsonb DEFAULT '{}'
) returns table (
  id bigint,
  content text,
  metadata jsonb,
  similarity float
)
language plpgsql
as $$
#variable_conflict use_column
begin
  return query
  select
    id,
    content,
    metadata,
    1 - (documents.embedding <=> query_embedding) as similarity
  from documents
  where metadata @> filter
  order by documents.embedding <=> query_embedding
  limit match_count;
end;
$$;sql

Hi there,

I'm contacting you from the Render support team.

We're getting a handful of support tickets into Render relating to deployments of Flowise following the guide at https://docs.flowiseai.com/deployment/render.

At Render we use ephemeral disks (https://render.com/docs/disks) so a SQLlite database written to ~/.flowise is lost when new deploys are made, instance types changed, etc etc and consequently Flows are lost. The video on the guide page says that using a starter plan or above will keep the Flows active and saved - It's true, the service won't sleep and provided nothing changes on the service then the flows would remain but any new deploys, or something that triggers the service to be restarted would mean the database is gone and flows lost.

Adding a disk to the service (only possible for starter and above instance types) mounted to /opt/render/.flowise would prevent this behaviour and store the sqlite DB in a persistent location. Ideally this path could be user configurable via a environment variable but that's a nicety.

We'd appreciate a call out in the documentation to highlight this behaviour and workaround.

I'm trying to install flowise on gcp following the instructions here - configuration/deployment/gcp.md

Successfully cloned the flowise repo to my local env (M3 apple silicon) here using

git clone https://github.com/FlowiseAI/Flowise.git

next

cd Flowise
pnpm install
pnpm build

on running pnpm build received an error

 WARN  Unsupported engine: wanted: {"node":">=18.15.0 <19.0.0 || ^20"} (current: {"node":"v22.0.0","pnpm":"9.1.0"})

[email protected] build /Users/georgeng/Documents/GitHub/Flowise
turbo run build

• Packages in scope: flowise, flowise-components, flowise-ui
• Running build in 3 packages
• Remote caching disabled
flowise-components:build: cache miss, executing ec79dc23b3f1e7dc
flowise-ui:build: cache miss, executing fe365f7e32feb03b
flowise-components:build:
flowise-components:build: > [email protected] build /Users/georgeng/Documents/GitHub/Flowise/packages/components
flowise-components:build: > tsc && gulp
flowise-components:build:
flowise-ui:build:
flowise-ui:build: > [email protected] build /Users/georgeng/Documents/GitHub/Flowise/packages/ui
flowise-ui:build: > vite build
flowise-ui:build:
flowise-ui:build: The CJS build of Vite's Node API is deprecated. See https://vitejs.dev/guide/troubleshooting.html#vite-cjs-node-api-deprecated for more details.
flowise-ui:build: vite v5.1.6 building for production...
flowise-components:build: [16:33:11] Requiring external module ts-node/register
flowise-components:build: [16:33:13] Using gulpfile ~/Documents/GitHub/Flowise/packages/components/gulpfile.ts
flowise-components:build: [16:33:13] Starting 'default'...
flowise-components:build: [16:33:13] 'default' errored after 118 ms
flowise-components:build: [16:33:13] TypeError: Cannot assign to read only property 'atime' of object '#'
flowise-components:build: at onFutimes (/Users/georgeng/Documents/GitHub/Flowise/node_modules/.pnpm/[email protected]/node_modules/vinyl-fs/lib/file-operations.js:258:27)
flowise-components:build: at FSReqCallback.oncomplete (node:fs:188:23)
flowise-components:build: at FSReqCallback.callbackTrampoline (node:internal/async_hooks:130:17)
flowise-components:build:  ELIFECYCLE  Command failed with exit code 1.
flowise-components:build: ERROR: command finished with error: command (/Users/georgeng/Documents/GitHub/Flowise/packages/components) /opt/homebrew/bin/pnpm run build exited (1)

flowise-components#build: command (/Users/georgeng/Documents/GitHub/Flowise/packages/components) /opt/homebrew/bin/pnpm run build exited (1)

Tasks: 0 successful, 2 total
Cached: 0 cached, 2 total
Time: 16.786s
Failed: flowise-components#build

ERROR run failed: command exited (1)
 ELIFECYCLE  Command failed with exit code 1.

Did I miss anything?

Thanks.

I am deploying flowise by following all the guide using the url https://docs.flowiseai.com/configuration/deployment/aws#setup

Got an error regarding ssl_version while running docker-compose up -d,

With the recent deployment of Version 1.6.5 did something change in the API response for the Prediction API end point that Zapier is leveraging? For all Zap steps for Predictions, they now error out with the following code:

Failed to create a prediction in FlowiseAI
Cannot read properties of undefined (reading 'sourceDocuments')

Hi everyone,
I using Flowise deployed into a Render environment and it was working with Pinecone for days. But today I deleted my index to start a new one from zero and it stopped working. Then I deleted my project and created a new one and then a I saw that it was created on a new free environment since I am not a paid user, then I thought that this could be the problem, but even with a new project deployed on free environment, its not working and I am receiving the following error:

ErrorWithoutStackTrace: PineconeClient: Error calling upsert: Error Without StackTrace: PineconeClient: Error calling upsert Raw: FetchError: The request failed and the interceptors did not return an alternative response

Hi,
I am using Render with a persistant disk (Starter plan: 512 MB / 0.5 CPU) and I have deployed a conversational agent with buffer memory, during training the application goes down once and then I can restart the training from where it stoped and conclude the deployment of the chabot. The training uses embeddings from OpenAI to embbed a PDF File of 9 MB (900 pages) into a Pinecone vector store.

But when I share the chatbot with a second user, the server goes down and I the training is lost too.
How can I solve this problem? Should I use a 2GB / 1 CPU server? Or I am doing something wrong in terms of configuration and deployment?