I have finally found some time to look at openldap stuff again, and have
been implementing the changes required to fix the final failures in
test058. But to my annoyance the race tests continued to fail :-( It
turns out that in addition to exercise the race condition outlined in
test058 it managed to fall into a stream of other race conditions.
These races are not related to the test058 configuration but to the
initiation of the syncrepl refresh phase.
The new test061 runs an extended version of the race test from test058,
but in a simplified configuration. So far I have found and (hopefully)
fixed 4 race conditions, and seen the glimpse of 2 or 3 more problems.
ITSes and fixes are coming, but as of now you should expect test061 to
fail occasionally.