On Thu, Mar 28, 2013 at 2:42 AM, Kristian Nielsen <knielsen@knielsen-hq.org> wrote:
Kristian Nielsen <knielsen@knielsen-hq.org> writes:
Pavel Ivanov <pivanof@google.com> writes:
Unfortunately with your recent commit (revision 3546) test rpl.rpl_gtid_startpos doesn't work when GTID patch is applied on top
I just pushed another patch (bzr revision 3548) with fixes for a similar failure in rpl.rpl_gtid_startpos. I still do not understand why it would be affected by "if (!is_relay_log && read_state_from_file())". However, the bug was a race in the test case, it failed or not at random depending on timing. So could it be that it just appeared to depend on this line, while in reality the issue was different timing when running the test?
It looks like revision 3548 indeed fixed the problem. It was very consistent difference in timing. :) Thank you, Pavel