Why User Defined Dashboard in not part of Git ?

[prabhakaran_mai]

I understand the difference between a User-defined dashboard and a LookML dashboard. UDD not part of git tracking is a big inconvenience for multi-instance deployments.

Within Looker, I am able to generate the LookML code for the UDD. Why can't Looker, behind the scenes, get the LookML code for UDD and make UDD part of git version control?

[Dr__Strangelove]

Git integration is reserved for "developers". There is a lot of complexity to version control, and having "regular users" use Git would be very challenging to them. We could potentially automate it fully, but there are usually only a few dozen to a few hundred LookML files under version control. We can easily have 10,000 UDDs in a system. That adds a lot to the version control system.

For multi-instance deployments, tools like Gazer https://github.com/looker-open-source/gzr are perfect for managing the movement of dashboards across systems.

[prabhakaran_mai]

@Dr__StrangeloveThanks for your response,

In multi-stage environment, where you have Dev > UAT > Prod . Only "Developers" have access to dev instance. If there is a system wide feature that can be enabled to track UDD (Behind the scenes , saves as LookML dashboard) , Only dashboards that Developer creates and saved into Dev "Shared" folder can be version controlled.

UDDs generated by business user (yes, it would lot of them ) in Prod are not version controlled. Also these dashboards don't need to be migrated.

I have used Gazer but it is an extra step, also if there is link between dashboards, it needs to updated with slug id for this to work. I prefer an out-of-box solution.

Regards,

Prabha Arivalagan

[Michael_Jassows]

Prabha,

  LookML Dashboards would seem to solve both the atomic release problem as well as the version control issue with UDDs.  However, the usability isn't quite on par with UDD's & Looks (no simple API access for data, all LookML dashboards in the same folder, no "LookML Looks").

I'm curious how large enterprises deal with this issue, or do they just take a potential down time for each breaking model change while they gazr in the updated dashboards?