Our CI sytem recently reported a failure in this test recently in October where the resource consumption metrics from an interrupted index build was higher than expected. One plausible explanation for this test failure is that there were write conflicts at the storage engine layer leading to the application layer retrying the index build sub operation, resulting in the operation metric incremented more than once (this is by design). If we are able to detect instances of write conflicts during the test, it would be helpful in investigating future occurrences of this failure scenario. This can be accomplished by increasing the verbosity level of the kWrite log component.