Product Feedback

  • Hot ideas
  • Top ideas
  • New ideas
  • My feedback
  1. Provide a working example of TensorFlow 2.0 subclassing API running in eager mode in a Pandas UDF

    Currently there is no example of working weight loading in a (possibly nested) subclassed TF2.0 model (tf.keras, the new standard) running inside a Pandas UDF, since this is the new industry standard of ML it would be great to have an example of how this is done in databricks

    1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  2. Awards Garage Door

    URL:
    https://www.get-awards.com
    Keywords:
    Garage Door Repairs, Garage Door Installation, Garage Door Springs, Garage Door Opner And Remote Repair And Installation
    Description:
    Garage Door Repairs and Installation.
    Address:
    11301 Coble Rd
    Charlotte, NC
    28227
    Phone:
    704-245-7272

    1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  3. Integrate Data Science Version Control (DVC) into Databricks Platform

    There is a special Version Control Framework for Data Science projects available which can be installed on top of GitHub etc. (https://dvc.org/). Would be great if we can leverage from this framework directly in Databricks.

    2 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  4. service principal usage while CI/CD instead of Personal token

    service principal usage while CI/CD instead of Personal token

    1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  5. we can provide relationships between Spark SQL tables

    Add the capability to provide informational relationships, constraints and hints that would help building a data model within Databricks itself. I understand this is a Spark feature which is pending since 2017 to provide Informational Referential integrity [SPARK-19842] but it hasn't moved on.If Databricks can provide a similar feature at least with Delta then there is no need of using an intermediary model-staging service like a RDBMS or OLAP or MPP when visualizing using tools like PowerBI or Tableau.

    21 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  6. Stop automatic scroll to bottom In logger panels

    You can stop automatic scroll to bottom In logger panels. It would help us looking into the logs. Right now as soon as we reach the relevant logs, it pulls us back to the bottom. This is very irritating.

    4 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  7. dbutils.fs.ls would return the lastmodified timestamp of files in addition to the size.

    The dbutils.fs.ls command returns the path, filename and size of the files it lists. With the timestamp the input files can be processed in the proper sequence.

    12 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  8. Historical Ganglia snapshot included more metrics

    It's nice that clusters have historical snapshots of the ganglia UI. It currently shows the "loadone" metric for all nodes on the cluster. It would also be nice to add some report metrics to this snapshot -- e.g. memreport, cpureport, networkreport, and disk_report. This makes it easier to debug issues with jobs that terminate clusters upon completion or failure.
    Thanks!

    1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  9. I had a searchable way to access workspaces instead of a non-scrollable menu.

    I will have many databricks workspaces for my clients to search through and manage, possibly 100+. Current list is not scrollable when you have many workspaces to juggle through. I've already run out of space with 8 workspaces. I'd like to search by DBX resource name & Azure subscription it's deployed in.

    3 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  10. Databricks could integrate better with a Python IDE

    Similar to the RStudio Server integration which has become availble recently, it would be great to have an integration with at least one Python IDE (PyCharm, Spyder, etc.). Is such an integration planned for the near future? The development of code in notebooks can be rather cumbersome for larger projects.

    26 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  11. ADLS AD integration

    When an Azure Data Lake Store Gen 2 is mounted on the cluster, I'd like users to be able to mount only the ADLS file systems folders that they have access through Active Directory.

    At the moment if a user mounts a folder from ADLS, that folder is visible on the whole cluster.

    3 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  12. Manage folder permissions through API

    I would like to manage access (permissions) to folders through the API instead of the GUI. Such that I can automatically enforce folder structure/access.

    45 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  13. 6 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  14. Databricks shaded all of its Java/Scala dependencies

    It'd be awesome if Databricks shaded all of its Java/Scala dependencies so that there is zero chance of version conflict in user-supplied libraries. Having just spent a huge amount of time researching why a library was failing and finding the problem is a ScalaPB version incompatibility, I would have very much appreciated this. If shading everything is a problem, maybe there is way to have the UI indicate 'hey, you loaded library XYZ which depends on ABC, which conflicts with our EFG, just FYI'.

    18 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  15. Jobs had "clean up" commands / cells that run on failure

    This can serve two purposes:


    • There may be some temporary data that is written that we'd want to clean up regardless of the job's success or failure, and this can ensure that occurs.


    • There may be some clean up that we want to occur only if the job fails. For example, if we partially write some data, the clean up could attempt to back that out.


    4 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  16. There was an email list for success notifications *only after a previous failed run*

    We use the email on job failure feature to notify business users when a job has failed, so they aren't wondering why their data isn't updated. However, sometimes the job succeeds on an automatic retry (which means the job failure notification is irrelevant) so we have them on the email on success list too.

    Unfortunately this means they get pointless notifications every day telling them that everything worked fine. It would be good if there was an email list that would only get notified when a previously failing job succeeded.

    3 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  17. It would be useful to be able to add a folder to git

    It would be useful to be able to add a folder to git

    7 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  18. Allow organizing jobs by folder

    Please add folder support for the jobs page.
    When you have a lot of jobs, it would be nice to group them by folder for organization purposes.

    15 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    2 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  19. Collate Large Quantities of Error Notifications

    When there's an issue running the code multiple error notifications can be generated, ore than can actually be shown at one time. This can result in several minutes of closing error notifications so all of a cell can be seen which makes databricks less fun to use. A solution could be if there are more than a certain number of errors (say 5), collate the extra errors into a single message e.g. "+20 more errors". Alternatively a Close All Errors button could be useful.

    4 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
  20. Add detach/reattach/reset notebooks from CLI

    It would be convenient to be able to detach/reattach/reset notebooks from the Databricks CLI.

    12 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Other  ·  Flag idea as inappropriate…  ·  Admin →
← Previous 1
  • Don't see your idea?

Feedback and Knowledge Base