Profiler - known issues

CloverDX Platform
1

Dear CloverETL Profiler users,

this topic describes known issues of CloverETL Profiler, especially during the beta testing phase.

Thank you for understanding,

Your CloverETL Profiler team.

2

Export to XML on Mac

platform: OSX
affected version: 0.9.0-beta1

On Mac, clicking on the “Export to XML” button in the Reporting Console does not ask to save the XML to a file, but the browser tries to display the XML itself. However it does not show the contents of the XML, just a blank page.

Workaround: the above behavior occurs only in the browser that is internal to the CloverETL Profiler. So to export the XML, open the Reporting Console in a new browser window via an action in the toolbar, please see attached screenshot. In the external browser window, the XML is saved to a file correctly.

3

Internal browser in some Linux distributions

The internal browser used on Linux needs XUL Runner (please see installation instructions). However, some Linux distributions provide only a too new version of XUL Runner (>=2.0). If you’re encountering issues with the internal browser, please make sure you have a supported version installed (1.9.X at most). In case you have multiple versions installed you can point CloverETL Profiler to the correct installation by adding “-Dorg.eclipse.swt.browser.XULRunnerPath=/path/to/xulrunner” (with the correct path for your system) to profiler.ini file.

4

Processing data sources with large number of fields(columns)

When processing data sources with many fields or columns (e.g. 200 and more), you might encounter the following error when running a job:

Exception in thread "main" java.lang.OutOfMemoryError: Direct buffer memory

To overcome this, please increase the available memory for the profiling job via the run configuration of the job (accessed via the Profiler > Run Configurations menu, see attached screenshots). For example, 250 fields usually require around 600MB of memory.

5

UPDATE: Processing data sources with large number of fields(columns)

The issue described in the previous post will be very rare in the upcoming beta 2 of CloverETL Profiler. The new version profiles data with a much more compact CloverETL graph. The only situation when the issue will occur is if you enable time consuming metrics (median, modus or unique count) on a very high number of fields (e.g. 200).