Managing Tasks with SnowDDL

[leopasta-enable]

Hi folks,
I am curious to know how other users of snowDDL manages task deployment? Do you suspend all tasks before deploying (in case snowDDL detects a change and needs to apply ALTER TASK)?
How do you know which ones to resume afterwards?

Interested in what patterns folks have used in the field

[littleK0i]

For serious deployments general recommendation is to avoid tasks in all cases and use modern orchestrator instead, e.g. Dagster. It helps to see a full picture and build super lineage across all data sources, transformations and reporing. Tasks are limited to Snowflake only.

For basic deployments it seems like a simple approach works well:

1. Let SnowDDL create & change stuff. Tasks are REPLACED in most cases, not ALTERED, so suspension should not be required.
2. After running "apply" and finishing all other manual SQL queries, run SHOW TASKS and find suspended ones. Manually "resume" such tasks. You may also trigger these tasks immediatelly.

You may write a procedure or a script to automate it.

It works well if number of tasks is relatively low (up to 50-100), if tasks are simple and if you do not have complex DAGs with tons of branching. When you start noticing complex DAGs - go for orchestrator as soon as possible.