Starting version 2025.4, there is a change in the behavior of multi-runs executing in the Cloud.
The purpose is to achieve a large increase in performance so that larger multi-runs can be processed rapidly.
Affected runs:
This change relates to multi-runs executing in the Cloud:
- Critical Link Analysis (CLA)
- Generalised Multi-run (GMR), which can be enabled for a number of run types
- Fire Flow - The Analysis tab of the GMR Configuration Object allows you to specify testable objects either from a selection list, or from the list of GMR test cases. This list of testable objects overrides the selection list in the fire flow options.
- Inside the Fire Flow Options, clear the selection from the "Include Nodes" selection list.
- The selection list is redundant in GMR but causes inefficiency. The role of the selection list is taken by the Analysis tab in the GMR Configuration object.
- When using the new cloud orchestration, go to GMR Options, below the GMR Configuration drop target set "Test cases" to "N". As general guidance, N = "Number of Tests" / 2000, but ensure that N >= 3.
- For cloud orchestration, this setting has an effect on performance for Hydrant Fire Flow GMR, but not other GMR types. A more suitable approach will be arranged in future.
On premise runs, using groups of agents, are not affected by this change.
Enhanced performance will be noted for larger multi-runs. Smaller multi-runs are less likely to experience dramatic improvements.
The new cloud multi-run workflow is enabled by default but you can choose to retain the previous behavior via the Enable Cloud Orchestrated Multi-runs check box in the Options - General dialog.
Changes in the desktop client
Most of the changes are hidden from the user. There are a few visible changes and some functional changes.
For multi-runs, the Job View window shows a single progress bar that deals with all test cases. Previously, there would be one progress bar for each cloud machine.
If you select the test case progress bar and go to the Job Progress view, you will see an updated count of the number of test cases queued, running and completed.
Technical note
The changes that have been made are as follows:
-
The queuing of sub-runs is done on the desktop client but the new method means that queuing is done in the cloud.
-
This reduces load on the desktop machine.
-
The number of cloud machines that can be invoked is raised above 80 to a substantially higher limit. (The new limit is kept under review.)
-
The use of machines in the cloud is more efficient.
-
-
Upload and download between the desktop client and the cloud has been rationalized.
-
Aspects of data processing have been improved, particularly in GMR.