You might have some luck using some APM tooling such as New Relic or Data Dog. They might offer a free tier that will provide enough info. It may point out where the system is spending time. It might pay to look at the processlist within MySQL in periods of slowness to see if there are any queries spending a long time responding to their request.

Hi....that is a fascinating perception. I review any significant changes in the inward functions of the test in 1.9. At any rate nothing that would majorly affect execution, simply bug fixes.

Goodness, then again, actually one of the progressions was MDL-15494, which made the test verify whether a blunder happened, as opposed to simply expecting that each call to refresh the information base succeeded and continuing indiscriminately. That may be uncovering a difficult that was formerly occurring, however not took note.

Any time you can spend looking around, and attempting to work out what is really happening would be significantly refreshing. This is difficult to duplicate in an advancement climate.