Agree on the limitations. For some reason a Databricks Workflow cannot contain more than 100 steps. Luckily there is now a new feature where a workow can contain a new kind of step which triggers another job. So now you can atleast subdivide you job into multiple smaller ones and then have a mster job that triggesr all the sub-jobs. But still, it would be way easier to just not have that limitation. It feels kinda artificial :/
@vyacheslavs56428 ай бұрын
You can use Terraform to provision your Workflows, Tasks, Clusters, Notebooks, etc. programmatically. Then Terraform scripts (*.tf, *.hcl) can be uploaded to Git and used in CI/CD as well.
@BryanCafferky8 ай бұрын
Thanks for your comment. Terraform is not open source anymore which causes me to pause on its future. OpenTofu is the new open source Terraform. You can also use Python with the Databricks Python SDK, or just Python with the Databricks REST API or the new Databricks Asset Bundles.
@palanikumar4150Ай бұрын
Can you please send any reference for creating Workflows. tasks and run notebooks using terraform?. So far I'm using terraform only to create databricks workspace and cluster.
@youfran Жыл бұрын
I wish they would add the possibility of adding workflow dependencies to other workflows. As a data engineer, you need this 100% of the time.
@BryanCafferky Жыл бұрын
Not sure what you mean. Could you elaborate?
@youfran11 ай бұрын
@@BryanCafferky I meant would be immensely helpful if Databricks workflows offered the feature to set a trigger mode based on the completion or state of other workflows, given we have the limit of 100 tasks per workflow.
@8aravindk11 ай бұрын
Hi @bryan, Why are these videos still not in the playlist on your website, it's been 2 weeks since you posted them here. I'm looking under the DataBricks Section and can't find them. I think your website should be first class citizen for locating your videos as well. Cheers and thanks for the helpful videos.
@BryanCafferky11 ай бұрын
Hi @Baravindk, They are in the YT playlist and the GitBook points you to the playlist rather than listing all the videos therein. To make new videos more easily found, I added a new videos menu to the GitBook and added these. These videos are in the KZbin Master Data Lakehouse playlist. Thanks
@michasikorski667111 ай бұрын
I have workflow with task A and task B, and 10 mores. I would like to have widgets or parameters like A : True, B : False... and it would decide if task should be skipped or now. Is it possible? How?
@afonso007810 ай бұрын
Thank you for sharing your knowledge! One question: is there a way to create this workflow using some type of ci/cd? for example, creating a development branch and pull request to merge in a master branch? The main idea is to create the workflow into a development environment and send it to the production environment.
@BryanCafferky9 ай бұрын
Yes. There are several ways. I am using the Databricks Python SDK from an Azure DevOps pipeline to do this. However, workflows are not stored in the repos so you'll need to use the UI, get the JSON and paste it into a file in your repo. learn.microsoft.com/en-us/azure/databricks/dev-tools/sdk-python You can also use the new Databricks Asset Bundles learn.microsoft.com/en-us/azure/databricks/dev-tools/bundles/
@datoalavista5817 ай бұрын
Brilliant !! Thank you so much
@BryanCafferky7 ай бұрын
You're Welcome!
@conconmc Жыл бұрын
Hi Bryan, wondering if you could a video of databricks and DBT? Would be interested in your thoughts :)
@BryanCafferky Жыл бұрын
I have not used dbt but from what I have seen it is very powerful. Thanks
@SaiKumar-ub6jo3 ай бұрын
Can you help how we can create the drop down for task parameters in worflow
@BryanCafferky3 ай бұрын
You use widgets. Doc here learn.microsoft.com/en-us/azure/databricks/notebooks/widgets
@SujeetKumarSinghlive9 ай бұрын
It helps lot , Thanks!
@BryanCafferky9 ай бұрын
You're Welcome!
@RajeshPhanindra11 ай бұрын
When creating a workflow, does it allow you to drag and drop tasks?
@BryanCafferky11 ай бұрын
No. The UI is more select and set the properties. The UI will update to the properties like dependencies.
@elprofesornet8897Ай бұрын
Hi Bryan, excellent material :) One question, would you still stick to workflows if you have to upload/download files using SFTP or import data consuming REST APIs?
@BryanCafferkyАй бұрын
Thanks. It depends on how the SFTP was being transferred. If some other external process dropped files to ADLS via SFTP, that seems to be fine. a Workflow could pick the file(s) up. If it needed to execute the SFTP to get the data within the workflow, This blog suggests using ADF to get the data to ADLS and then the workflow could do the rest. Notebooks can call REST APIs so it is possible to get data using REST APIs but whether it works in a given scenario would need to be evaluation by the data engineer.
@shankhadeepghosal7314 ай бұрын
how to use if else branch logic ?
@manasr39695 ай бұрын
love this video. The dashboard refresh is supercool
@MariusS-h2p6 ай бұрын
19:25 That's not a future option, that's just the category?!
@lukasu-ski43254 ай бұрын
Yep :) thought the same thing
@Databricks Жыл бұрын
Great summary!!
@zoji95667 ай бұрын
Invaluable. Thank you 🙏
@joshuatrampier435510 ай бұрын
How do you delete a task from a workflow?
@BryanCafferky10 ай бұрын
click on the task in WF editor and click on the trash can.