Hey all, Not sure if it's our specific setup, but I wanted to flag that spicedb v1.45.4 has been breaking our builds. Pinning spicedb to v1.45.3 fixes the issue, but I couldn't see anything suspicious in the release notes. I am also running

authzed/spicedb:v1.45.4

in my local docker-compose setup without any issues. For reference I attached the startup log output. When my application is trying to establish a grpc connection with spicedb all I see is timeouts. https://cdn.discordapp.com/attachments/844600078948630559/1417869914521206794/message.txt?ex=68ccb682&is=68cb6502&hm=365eecfb402485e6aa5a373a79a1d260b7b107885780f6034fefaad6bcc954fc&

how are y'all running spicedb? are you using the operator?

Ah sorry, I meant to attach the github actions file. This is happening in our github testing pipeline.

jobs:
    build:
      steps:
        - uses: actions/checkout@v4
        - name: Install Native dependencies
          run: apt update && apt install -y build-essential curl
        - name: Install SpiceDB
          run: |
            curl -sS https://pkg.authzed.com/apt/gpg.key | gpg --dearmor --yes -o /etc/apt/keyrings/spicedb.gpg
            echo "deb [signed-by=/etc/apt/keyrings/spicedb.gpg] https://pkg.authzed.com/apt/ * *"  | tee /etc/apt/sources.list.d/authzed.list
            chmod 644 /etc/apt/sources.list.d/authzed.list
            apt update
            apt install spicedb
        - name: Start SpiceDB
          run: spicedb serve-testing &

ah, gotcha. just out of curiosity, have you seen/tried the action-spicedb action? https://github.com/authzed/action-spicedb

I have not, but this looks like it could've saved me an afternoon on the initial setup haha 🙂 I think the first test still manages to connect, but all subsequent calls fail to connect, so that first call might actually crash the server? Here's the relevant log entries (we're running Elixir using [authzed_ex](https://github.com/goodhamgupta/authzed_ex) as client)

[warning] Received unexpected response when fetching SpiceDB schema diff: {:error, %GRPC.RPCError{status: 13, message: ":stream_error: :closed"}}
[error] My.App.Server #PID<0.638.0> received unexpected message in handle_info/2: {:gun_down, #PID<0.639.0>, :http2, :closed,
[#Reference<0.4109527196.2503213057.89019>]}

that helps. do you know what that first call is, or is it dependent on the ordering of tests etc?

It's a

DiffSchemaRequest

. Basically when starting up our server fetches a diff against the currently deployed schema and then figures out based on that which relations need to be deleted before writing the new schema and which entities to sync after the new schema has been written. This is the sequence:

with {:continue, state, diffs} <- diff_schema(state),
         {:continue, state} <- purge_relationships(state, diffs),
         {:continue, state} <- migrate(state),
         {:continue, state} <- sync_relationships(state, diffs) do

ahhh yeah. i've run into the same, but it seemed like a limitation of github actions

Just checked on my mac and the latest release is running without issues there. Trying to get some more logs out of my github action now

let me know what you find!

Ok, when running the

spicedb serve-testing &

command in the same step/shell as the test process everything works as expected (but I'd have to disable spicedb logs). If I start the spicedb instance in the step before starting the tests I see the same issue I described again.

that's bizarre

I do see a couple tests that managed to get past the schema migration though (and fail a bit later on when trying to write relationships)

wait yeah

Well it seems like it initially manages to connect as expected and run a couple commands against spicedb, but pretty quickly after that it seems like spicedb stops responding.

hmm

I tested again using

JarvusInnovations/background-action@v1

and this time it worked with v1.45.4.

- name: Start SpiceDB
        uses: JarvusInnovations/background-action@v1
        with:
          run: spicedb serve-testing --http-enabled &
          wait-on: http-get://localhost:8443/healthz

maybe its prematurely closing the binary

Could be, but I do wonder what caused the behavior to change 🙃 Anyway, using some ideas I took from the background-action I came up with the following github-script which seems to work 🤷‍♂️

- name: Start SpiceDB
        uses: actions/github-script@v8
        with:
          script: |
            const spawn = require('child_process').spawn;
            const spawnOpts = { detached: true, stdio: 'ignore' };
            const run = 'spicedb serve-testing &';

            const child = spawn('bash', ['--noprofile', '--norc', '-eo', 'pipefail', '-c', run], spawnOpts);
            console.log('SpiceDB started');

that's bizarre 🤔

Sure! https://github.com/authzed/spicedb/issues/2573

thank you!

Seems like adding

--log-level=warn

fixes the issue as well and I can go back to a plain shell command :tada https://github.com/authzed/spicedb/issues/2573#issuecomment-3311738554