Sunday night queue component issues

We had some issues in one of our components (job handling queue) between 10pm and 10:15pm PST on Sunday night (03:00–03:15 UTC Monday). That had resulted in some failed orchestrations (scheduled at that time or starting it's tasks at that time) . We're planning to upgrade the component in the coming weeks, but if the error occurs again, we'll upgrade it immediately. The upgrade will be accompanied with a short maintenance downtime.


We're sorry for any inconvenience.

Transformation Overview

We have slightly changed the layout of the transformation detail (sections moved to tabs) and added an Overview graph

The graph shows the data flow around the current transformation, including all required transformations, input/output tables and transformations from the same bucket, that work with any of the tables as the current transformation. This allows you to get a bird's eye view on what you're working with. 

The difference between this and Visual SQL (or SQLdep) is that Visual SQL goes deeper in the transformation and analyses the queries, whereas Overview shows you the environment around the transformation. 

GoodData LDM Visualizer

There's a direct link to GoodData LDM Visualizer in the model page in GoodData Writer. So in a single click you can compare what is defined in Keboola Connection and what has been already uploaded to GoodData and how the model is interpreted.

Transformation Descriptions

We're now showing bucket and transformation descriptions in the UI. Currently there's no way to change the bucket description - it is defined setup when the bucket is created, but we're working on a way to make them editable. Transformation descriptions can be changed easily in the transformation detail

Editing in Storage Console

Due to a possible significant inconsistency in real and estimated number of rows (and table sizes) on the MySQL Storage backend which could lead to data loss when editing the data sample, we turned off the data sample editing for all IN and OUT MySQL buckets. SYS buckets are fully editable and will always show and edit all available rows. 

Redshift works as expected - editing is allowed for all tables if they contain less than 800 cells (excluding headers). 

Transformations EA Preview

We'll be releasing updated version of the Transformation API (and accordingly modified UI) later this week. Changes will include:

All transformation and sandbox job are asynchronous. 

More reliability, durability and scalability of both UI and API. 

You can monitor all transformation jobs in the Jobs app, processing jobs will show detailed 

Running a transformation or a bucket will not keep the modal window open. 

It will provide you a link to the job detail instead. 

Sandbox jobs will keep the window with the details, though.

Custom credentials are deprecated.

We're providing you with enough power to process your data. If you want to run transformations on your own database servers, please contact support@keboola.com.

Running disabled transformation is disabled.

May sound weird, but it is. You were allowed to run disabled transformation from the transformation detail in the UI. 

If you want to separate a transformation, set it to a different phase or migrate it to another bucket.

Testing period

If you want to make sure everything works fine in your project, you can try it out. The new UI is available and is connected to the new version of API in the Applications app.

You can try running certain transformations, create sandboxes etc. If you want to try the new API within an orchestration, you need to change manually the component name in the configuration table (sys.c-orchestrator.*) from transformation to transformation-new.

In the Orchestrations UI the task name will show Transformations (EA Preview)

When your testing is done, please reset the value back to transformation. 

Stay updated for the release date and please report all bugs and concerns to support@keboola.com.


Transformation Events

As a part of the ongoing Transformation API overhaul we've changed transformation events. We tried to keep it simple, so there's one event for each:

  • Engine startup
  • Start of a phase
  • Database cleanup (happens at the beginning and end of a phase)
  • Input mapping
  • Input mapping that takes longer than 120s (configurable)
  • Transformation (as a whole, not a query in the transformation)
  • Query that takes longer than 120s (configurable)
  • Output mapping
  • Engine shutdown (success or error)
There are no more any START or END events and the engine produces less events. Further activity can be found in related storage events or jobs (eg. table imports and exports).