joy_timmermans
11/15/2020, 2:50 PMError: No such command "bash". and the file i’m using is https://gitlab.com/meltano/meltano/blob/master/docker-compose.yml. doesn someone have any ideas?emilie_schario
11/15/2020, 2:51 PMjoy_timmermans
11/15/2020, 3:12 PMdouwe_maan
11/16/2020, 4:42 PMdocker-compose.yml from the docker-compose file bundle: http://meltano.com/docs/ui.html#docker-compose https://gitlab.com/meltano/files-docker-compose/-/blob/master/bundle/docker-compose.yml
also when i tried to use meltano commands in docker to create docker compose file it didnt create it but that is prob because it created in the docker container not the directory i was inDid you have your project directory mounted when you added the file bundle? It would look something like this:
docker run -v $(pwd):/project -w /project meltano/meltano add files docker-compose
That would add docker-compose.yml to your current working directory (pwd), which is mounted into the container at /project .joy_timmermans
11/16/2020, 5:04 PMdocker-compose run --rm meltano --version for example? adding a database container i can do. but why not give one environment that does it all out of the box?douwe_maan
11/16/2020, 5:14 PMdocker-compose.yml gives you an environment appropriate for development, where any changes you make in the UI end up in meltano.yml in your project dir, and where a system database is not necessary.
For a more production-like setup, there's https://meltano.com/docs/containerization.html#docker-compose and https://gitlab.com/meltano/files-docker-compose/-/blob/master/bundle/docker-compose.prod.yml, which uses the project-specific Docker image that doesn't allow any changes to meltano.yml anymore (since they wouldn't be persisted outside the container), and uses a separate system database.
In both setups, the service is called meltano-ui rather than meltano (so you'd need docker-compose run --rm meltano-ui --version) to make it clear that the UI does not represent all of Meltano, and that that service is not responsible for actually running Meltano ELT pipelines, which is where the airflow-scheduler service comes in.
I think in your case, it would be easier to have a single meltano service that runs both the UI and the scheduler, that could then easily be used through docker-compose run --rm meltano <args> . This would require a way to tell meltano ui to also run the orchestrator/scheduler, using an env var or CLI argument. This was recently discussed in https://meltano.slack.com/archives/CFG3C3C66/p1603820355263900?thread_ts=1603627483.239000&cid=CFG3C3C66, if you'd like to have some more context.
If you think this would be useful in Meltano and simplify things for you and new users, do you think you could create an issue? 🙂joy_timmermans
11/16/2020, 6:55 PMdouwe_maan
11/17/2020, 4:22 PMwhat does the scheduler do then?The scheduler is the part of the orchestrator (Airflow in our case) that's responsible for automatically running the scheduled pipelines at the defined intervals: https://meltano.com/docs/orchestration.html If you're not using an orchestrator and its scheduler, you have to manually set up some system (like cron) to automatically run
meltano elt <args> whenever you want it.douwe_maan
11/17/2020, 4:24 PMAlso that you say there is no system database necesary how do you create a dashboard then?The system database only stores a few environment-specific types of data: https://meltano.com/docs/project.html#system-database Dashboards created using the UI are stored inside the project, in the
analyze/dashboards directory, so they don't require a system database.
you need a extractor and a leader? but the loader is a database?The loader is a plugin responsible for loading data into a database, yeah, and the UI will try to use the loader's configuration to automatically connect with the same database so that you can use it to run queries against the loaded data.
douwe_maan
11/17/2020, 5:10 PMI really like what the project could bring but it seems a little hard to grasp at first for a new user.You're not the first to share that feeling 🙂 Have you seen https://meltano.com/docs/getting-started.html? It may not have been around yet when you first started playing around with Meltano, but I hope it does a decent job of introducing a user to Meltano and its concepts. We've also got 2 open issues to make it easier to follow that may have helped a new user like you as well: https://gitlab.com/meltano/meltano/-/issues/2383 https://gitlab.com/meltano/meltano/-/issues/2382
i think what would really help is making a small youtube vid where you setup Melton from a clean install to a dashboard and explain some of the things why it is that why along the way. just a short 5-10 min to get new people going.The Getting Started guide and most of the docs are focused on the CLI and EL(T), since the UI and the Analysis functionality (dashboards, reports) are explicitly not currently a priority, and the UI lacks various features of the CLI: https://meltano.com/docs/ui.html#current-status. So while a YouTube video could help users get started with the UI and dashboards, we don't actually want to focus too much on that side of Meltano, and would rather prioritize making it easier to use and get started with the CLI and EL(T) functionality. Would that work for you, or are you specifically interested in Meltano for its built-in analysis functionality as well?