[SOLVED] Function always returns "Execution timed out." on self-hosted Appwrite

I've been trying to get the Appwrite API to work in a Function, but I just can't seem to get it.

The function runs fine when I use just the Stripe API code, but once I start adding the Appwrite API to save some user preferences, it returns Execution timed out..

What I checked:

• I'm instantiating a new Appwrite client
• I'm setting endPoint, setProject, setJWT to the Appwrite client
• I'm creating the Appwrite Account client
• The Function initially had Users execute access, but I tried Any, to no avail

When I log the client

My /src/lib/appwrite.ts:

My main.ts:

Any hint on what I'm doing wrong would be appreciated.

Some findings:

• I'm using a custom domain. When I set the endpoint in the function to the non-custom domain (e.g. https://appwrite.xxxxxx.com, it's contining the function instead of just returning nothing
• In above code I'm using the Account SDK with a JWT, I've switched to the Users SDK with a global API key

More notes:

• Appwrite 1.4.3
• Self hosted on Digital Ocean
• I'm triggering this function from a Next.js frontend with createExecution
• The API key that I'm using never shows that it's accessed in the Appwrite dashboard, which is weird, because in the Function I initialise specifically with that key

When I get ECONNREFUSED, is that a server config error, or does it mean my credentials are invalid?

connect ECONNREFUSED 127.0.1.1:443
I faced this exact issue, when I setup Appwrite on something other than 443.

Since you're self-hosting, can you tell me what your traefik is set up like? Specifically the traefik service:

services:
  traefik:
    image: traefik:2.7
    container_name: appwrite-traefik
    <<: *x-logging
    command:
      - --providers.file.directory=/storage/config
      - --providers.file.watch=true
      - --providers.docker=true
      - --providers.docker.exposedByDefault=false
      - --providers.docker.constraints=Label(`traefik.constraint-label-stack`,`appwrite`)
      - --entrypoints.appwrite_web.address=:80
      - --entrypoints.appwrite_websecure.address=:443
    restart: unless-stopped
    ports:
      - 5080:80
      - 443:443
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - appwrite-config:/storage/config:ro
      - appwrite-certificates:/storage/certificates:ro
    depends_on:
      - appwrite
    networks:
      - gateway
      - appwrite

As you can see, traefik is set up to map 443 to 443.

If it helps, I am using Nginx Proxy Manager in front of Appwrite as a load balancer.

Do you know why the IP is 127.0.1.1?

ooh yeah that's weird

@safwan I came across the question you posted before, but figured it must be something else since I'm using the default ports.

This is what my Traefik service looks like:

Maybe a bit more information regarding my setup will help.

@Steven I have no idea why the IP is 127.0.1.1. Can I debug or log something that gives an answer to that question?

My setup:

• I'm running Appwrite on a subdomain: https://appwrite.xxxxx.com

• I've added a custom domain for the web to interact with: https://backend.xxxxx.com

• The Function variables:

  • setEndpoint: https://appwrite.xxxxx.com/v1 (I tried the custom domain https://backend.xxxxx.com too)
  • setProject: xxxxx (confirmed in Appwrite)
  • setKey: xxxxx (new generated API key with all access)

• The Function is deployed manually with Appwrite CLI. Build etc. is fine:

  • execute: users
  • I've rewrote it to just use updatePrefs and then getPrefs and then log it in the console.

This is the updatePreferences function:

I call it like so:

It always throws: connect ECONNREFUSED 127.0.1.1:443

Relevant parts of my .env:

Might've found something:

• When I run docker exec appwrite vars, the _APP_FUNCTIONS_RUNTIMES_NETWORK value is empty
• When I SSH to the server and check the .env file, the value is runtimes
• When I run docker compose up -d, the env vars don't seem to reload / update

I guess my Appwrite install is corrupt. I installed 1.4.2 and followed the docs to update to 1.4.3, then ran the migration. I'll try a 1.4.3 clean install.

Also strange:

cat .env | grep RUNTIME gives me:

docker exec appwrite vars | grep RUNTIME gives me:

Is this expected or is there some sort of .env cache? This is after changing the .env file and running docker compose up -d.

This might all go back to running the migration from 1.4.2 to 1.4.3. and the migration script breaking on the GitHub integration SSH key _APP_VCS_GITHUB_PRIVATE_KEY, because it's multiline.

I've manually updated the .env file after the migration from 1.4.2 to 1.4.3, but it seems something broke there.

Allright. I must be doing something simple very wrong, it's still not working** after a fresh 1.4.3 install.**

I don't get connect ECONNREFUSED 127.0.1.1:443 anymore, but the function times seems to stop running and times out right after I call any of the Appwrite functions, specifically updatePrefs.

Any idea on how to debug this?

docker logs appwrite-worker-functions gives me:

docker logs appwrite-executor-********-65114879c1f66e6288ad gives me:

docker logs appwrite gives me:

Nothing special.

cat .env | grep RUNTIME gives me:

docker exec appwrite vars | grep RUNTIME gives me:

The function code (node-20):

The appwrite service code in the function

what is APPWRITE_ENDPOINT now and where is your appwrite instance installed?

APPWRITE_ENDPOINT is https://appwrite.xxxxx.com/v1

My Appwrite instance is installed on a Digital Ocean droplet and publicly accessible. Path is /root/appwrite. I can share the domain / IP if you'd like to.

I created a new droplet yesterday with a Docker image pre-installed (link), then ran the Unix install command for version 1.4.3, found in the Appwrite Self Hosted docs

Side note, I'm triggering this function from a localhost:3000 next.js app with executeFunction after I created a magicUrl account and logging in.

The Appwrite client in the client side connects to the configured custom domain (https://backend.xxxxx.com)

Use case for this Function and accessing the Appwrite user preferences is:

1. Get all the user preferences in the Function
2. If there is not a Stripe ID saved in the Appwrite user preferences, call the Stripe API without an ID and let Stripe create one
3. If there is a Stripe ID saved in the Appwrite user preferences, call another Stripe API with the ID that we saved previously

Good lord, it's solved. The function and Appwrite install was fine.

I checked out the Digital Ocean one-click setup for Appwrite repository and there's a script that adds port 80 and 443 to the UFW rules.

The firewall blocked my Function from accessing Appwrite without throwing.

For anyone going through the same and doesn't want to have gray hair:

Check your firewall rules in the droplet:

If it doesn't show 443 and port 80 like this:

Update the firewall rules by running:

Double check if the rules are active now:

[SOLVED] Function always returns "Execution timed out." on self-hosted Appwrite

Hindsight: I wanted to install a fresh 1.4.3, that's why I didn't use the one-click Appwrite on Digital Ocean. But I see it's now always using the latest version.

If there is not a Stripe ID saved in the Appwrite user preferences, call the Stripe API without an ID and let Stripe create one

btw...i'd be wary of saving the stripe ID in user preferences because the user is freely able to update user preferences 👀

Thanks Steven, good tip.