Dependency resolution steps run many times, slowing synchronization (4209297)

A synchronization takes longer to run than would be expected based on the number of objects being synchronized. In the synchronization log, there are messages such as "Executing synchronization step (Group Phase #2)" and "Retrying failures of step (User Phase #2)". The repeated Phase #2 steps seem to be slowing down the synchronization. Why are they repeating?

This is the expected behaviour of the dependency resolution process, as described in the Target System Synchronization Reference Guide.

"Dependencies can arise between schema classes that require synchronization steps to be repeated. For example, object references cannot be set until the reference object has been added. Dependencies can also arise between schema properties within a schema class."

These dependency resolutions are what Phase #2 in a synchronization is resolving.

Each synchronization project has a setting called "Failure handling mode". This setting determines what to do if objects in a synchronization fail. By default the failure handling mode is set to "Retry". This setting increases the likelihood that a synchronization will complete successfully for all objects. However, it can also increase the length of time that a synchronization takes to complete depending on how many steps must be repeated. In order to improve performance, the failure handling mode can be changed to "Ignore". This can improve the speed of a synchronization, but the drawback is it increases the probability of uncorrected errors, leading to an incomplete synchronization.

To change this setting in the Synchronization Editor:

• Select "Database" | "Settings" from the Menu

• Under the "General" tab, enable the "Enable Expert Mode" checkbox

• Under the "Start Up Configurations", click "Edit"
• Select the "Advanced" tab, and the "Failure handling mode" is available

The two options available are:

• Repeat (Default). The Synchronization is repeated until the failed objects do not yield any changes.
  • Advantage: This results in a full synchronization. Certain, highly complex hierarchical structures cannot only be synchronized in this mode.
    
  • Disadvantage: Synchronization in this mode takes considerably longer if a failed object arises.
• Ignore. Where the synchronization ignores failed objects.
  • Advantage: Synchronization does not take longer because of failed objects that cannot be corrected by repeating (data error).
  • Disadvantage: Errors, which can occur through complex data dependencies, are no longer corrected. The synchronization results may therefore not be complete.

NOTE: Repetitions of failed objects means a reduced synchronization.

It makes sense to ignore failed objects if synchronization with this start up configuration is run frequently because the failed objects are processed again by the next synchronization. This also achieves full synchronization results and at the same time reduces the runtime of each synchronization.

For more information on the failure handling modes, please see the Target System Synchronization Reference Guide.