docker: 'compose' is not a docker command. about llama-gpt

[nmfretz]

Hi @arthurwolf, it looks like you may be facing one of two issues.

1. You don't have Docker Compose installed. If that's the case then you'll first need to install it before you can run docker compose up. To find out if you have Docker Compose installed, run docker compose --version in your terminal. If a version number is returned then you have it installed.

2. You have Docker Compose Standalone installed, in which case the syntax you should use is docker-compose up. Note the hyphen. To find out if you have Standalone installed, run docker-compose --version

Let me know if you get it sorted!

[arthurwolf]

╭─arthur at aquarelle in ~ 23-08-24 - 18:26:48
╰─○ docker compose --version
Docker version 20.10.25, build 20.10.25-0ubuntu1~23.04.1

Looks like that worked, thanks!

╭─arthur at aquarelle in ~/dev/ai/llama-gpt on master✔ 23-08-24 - 18:27:46
╰─⠠⠵ docker-compose up                                              on master|✔
Creating network "llama-gpt_default" with the default driver
Pulling llama-gpt-api-7b (ghcr.io/abetlen/llama-cpp-python:latest)...
latest: Pulling from abetlen/llama-cpp-python
14726c8f7834: Pull complete
7d676dc8a994: Pull complete
e9e2f08137fc: Pull complete
6cc9fece80c8: Pull complete
264b6a89a95d: Pull complete
50872ae560f9: Downloading [====>                                              ]  12.28MB/140MB
5ab61a038675: Download complete
4f4fb700ef54: Download complete
124cd1deda15: Download complete
23e537c28460: Download complete
c904885da53b: Download complete

Might be worth some kind of docs edit? Not sure how rare my situation is.

[rsebi]

╭─arthur at aquarelle in ~ 23-08-24 - 18:26:48
╰─○ docker compose --version
Docker version 20.10.25, build 20.10.25-0ubuntu1~23.04.1

Looks like that worked, thanks!

╭─arthur at aquarelle in ~/dev/ai/llama-gpt on master✔ 23-08-24 - 18:27:46
╰─⠠⠵ docker-compose up                                              on master|✔
Creating network "llama-gpt_default" with the default driver
Pulling llama-gpt-api-7b (ghcr.io/abetlen/llama-cpp-python:latest)...
latest: Pulling from abetlen/llama-cpp-python
14726c8f7834: Pull complete
7d676dc8a994: Pull complete
e9e2f08137fc: Pull complete
6cc9fece80c8: Pull complete
264b6a89a95d: Pull complete
50872ae560f9: Downloading [====>                                              ]  12.28MB/140MB
5ab61a038675: Download complete
4f4fb700ef54: Download complete
124cd1deda15: Download complete
23e537c28460: Download complete
c904885da53b: Download complete

Might be worth some kind of docs edit? Not sure how rare my situation is.

The expectation is that you know docker & docker compose. But a slight edit to the doc would be appreciated as pointed.

[arthurwolf]

Thanks a lot for all the help.

(I'll make a PR with the docs change)

Ok so it works now, but I have a second issue, asking here just in case it's easy, I'll make a new issue if not.

so I started with

docker-compose up

It worked.

So I wanted to try 13b next, so I did:

docker-compose down

Which worked, and then:

docker-compose -f docker-compose-13b.yml up 

But this gets me:

╰─⠠⠵ docker-compose -f docker-compose-13b.yml up                                                                                                                                                                      on master|✔
Starting llama-gpt_llama-gpt-api-13b_1 ... 
Starting llama-gpt_llama-gpt-api-13b_1 ... error

ERROR: for llama-gpt_llama-gpt-api-13b_1  Cannot start service llama-gpt-api-13b: network c6ea4bb425bd91c6090bfde1ef50c974ada03f2bfba27caff216184c6684a5d1 not found

ERROR: for llama-gpt-api-13b  Cannot start service llama-gpt-api-13b: network c6ea4bb425bd91c6090bfde1ef50c974ada03f2bfba27caff216184c6684a5d1 not found
ERROR: Encountered errors while bringing up the project.

Which I'm not sure what to do about...

The expectation is that you know docker & docker compose

I use docker a few dozen times a year, I've even made a few containers, but I'm not sure what's the criteria for "knowing" docker, this is my first time running into this one.

[arthurwolf]

PR created, see: https://github.com/getumbrel/llama-gpt/pull/63/files

[maxnatt]

As I mentioned in #81, Compose V1 is officially EoL and I think a proper solution would be simply to advice users to ensure they are using Compose V2: https://docs.docker.com/compose/install/