I'm running the latest version of the BVT (E55769-04) over OBIEE version 126.96.36.199.160719 on Windows 10 and getting some unexpected behavior from the UI Plugin, specifically FindDashboardSnapshotDifferences. We're trying to use the BVT to automate regression testing of a subset of our dashboards spread through our catalog, so we're feeding it a long list of prompted links instead of specifying values for the CatalogRoot parameter (which is set to an intentionally empty folder).
The list has about 194 prompted links spanning 25 dashboards.
BVT always says "All 0 dashboards and 194 dashboard pages loaded and snapshots taken successfully!!!", but then my -compareresults runs are showing that over half of these dashboard pages land in the "Failed to Download Tests" category. Looking in the log file, even though all the links are processed, there's no indication that the missing dashboard pages were actually navigated to. Since the lines in the log file are timestamped, I can also see that no time passes between the line "Checking for reports with load errors and retrying loading" and "All snapshots taken successfully", which points to no checking or reloading actually taking place.
Close examination of the log files suggests that the BVT isn't initializing enough browser instances for some reason. If I have ThreadQueueSize = 1, then 2 browser instances are initialized. If ThreadQueueSize = 3, then only 4 browser instances are initialized, and if ThreadQueueSize = 13, then only 14 instances are initialized, etc. Since each instance only processes 3 or less dashboards, these aren't enough to cover the 25 dashboards I need to have processed.
I did try to work around this issue by setting ThreadQueueSize to 25, but that caused its own problems.
Any help or knowledge regarding this problem would be appreciated.