Hello, is it possible to use tags to filter solids? I'm using DBT and it would be wonderful to be able to filter solids with something like

tag:key=value*

We don’t have this currently, but it is an excellent idea and something we are actively considering. In the interim we do have substring matching in the pipeline definition view in the upper right which grays out the solids that do not match.

Thank you very much @schrockn. I take the opportunity to congratulate you and the team for this incredible tool you're building 💯

Hey, Flavien. I’m not sure about your specific use-case, but I’ve had success defining separate solids for specific subsets of dbt models (i.e. for a specific set of dbt tags), and then using these to compose various pipelines. I’d be interested to know if there are reasons why this might not work for you and if there’s any way I can improve my approach.

Hey Ben, I was thinking of using them in an ad hoc fashion for example to backfill downstream dependencies of a given source in a bigger pipeline. Do you rely on DBT

manifest.json

to build your DAGs? I'd be very interested to have your feedback on working with both tools 😁

For now, I’ve just been using

dagster_dbt.dbt_cli_run

to manage my pipelines (and using the

models

argument to run a subset of dbt models). Very much at the beginning of my dagster journey (currently working on migrating a bunch of cron jobs to a more structured set of pipelines), so not sure how useful my feedback would be, but I have been enjoying it so far!

In my case I would like to take the same approach as mentioned in this article but with Dagster instead: https://www.astronomer.io/blog/airflow-dbt-1

Ah, I’m not (yet?) at a scale where the issues they are having with dbt are an issue for me. I’m a little surprised that the issues they have with re-running individual models can’t be worked around by using incremental materialisations, for example. But I’m not working at scale, so perhaps I’m just being ignorant/naive. Moreover, with dagster_dbt, the models are tracked as assets. This means you can run your selection of dbt models with the CLI but get metadata & monitoring for each model.