)]}'
{"test/unit/common/ring/test_builder.py":[{"author":{"_account_id":7233,"name":"Matthew Oliver","email":"matt@oliver.net.au","username":"mattoliverau"},"change_message_id":"e57d2ff4d16ecdb1cfd34dba26cd0343cd0f84ff","unresolved":false,"context_lines":[{"line_number":4497,"context_line":"            \u0027127.1.2.2\u0027: 413,"},{"line_number":4498,"context_line":"            \u0027127.0.0.3\u0027: 410,"},{"line_number":4499,"context_line":"            \u0027127.0.0.4\u0027: 410,"},{"line_number":4500,"context_line":"            \u0027127.1.1.5\u0027: 402,"},{"line_number":4501,"context_line":"        }"},{"line_number":4502,"context_line":"        self.assertAlmostPartCount(_partition_counts(rb, \u0027ip\u0027), expected,"},{"line_number":4503,"context_line":"                                   delta\u003d5)"}],"source_content_type":"text/x-python","patch_set":1,"id":"7f96bb07_9daeb82c","line":4500,"updated":"2018-01-12 01:36:47.000000000","message":"So even with this change I still get regular failures. Max difference I\u0027ve seen is a delta of 7 on 0.0.5 (from the above).\n\none way to solve it would be to set a seed (and hopefully find one what works for both py2 and py3) in the rebalance on line 4493.\n\nAnother option it to increase the delta. Given a partpower of 8, 5 servers and 8 replicas. the average (not taking tier dispersion) is about 409 parts per server:\n\n  pow(2,8) * 8 / 5\n\nThe current delta 5, gives us a range of 10 (+-5 from expected), which is 2.5% of this 409 number.\n\nIf we up the delta number to say 7, which is a range of 14 values (+- 7 from expected) which is 3.5% of 409, then this covers all the values I\u0027ve observed.\n\nI guess the question is, where did a delta of 5 come from. And is anything outside of that considered a problem? If so, we consistently hit outside this. If we\u0027re playing a percentage games, covering an extra percent per node (from 2.5% to 3.5%) I think is enough to cover the worst I\u0027ve seen. That would be changing the delta from 5 to 7.\n\nOr if we want a little more wiggle room in case there is the possibility of higher then 7. then a delta of 8 is 4% ;)","commit_id":"eea45a153aa8ac959bb4deae4ba3596ef62add06"},{"author":{"_account_id":7233,"name":"Matthew Oliver","email":"matt@oliver.net.au","username":"mattoliverau"},"change_message_id":"e57d2ff4d16ecdb1cfd34dba26cd0343cd0f84ff","unresolved":false,"context_lines":[{"line_number":4500,"context_line":"            \u0027127.1.1.5\u0027: 402,"},{"line_number":4501,"context_line":"        }"},{"line_number":4502,"context_line":"        self.assertAlmostPartCount(_partition_counts(rb, \u0027ip\u0027), expected,"},{"line_number":4503,"context_line":"                                   delta\u003d5)"},{"line_number":4504,"context_line":"        report \u003d dict(utils.dispersion_report("},{"line_number":4505,"context_line":"            rb, r\u0027r\\d+z\\d+-[^/]*$\u0027, verbose\u003dTrue)[\u0027graph\u0027])"},{"line_number":4506,"context_line":"        counts \u003d {k.split(\u0027-\u0027)[1]: d[\u0027placed_parts\u0027]"}],"source_content_type":"text/x-python","patch_set":1,"id":"7f96bb07_fd625c16","line":4503,"range":{"start_line":4503,"start_character":41,"end_line":4503,"end_character":42},"updated":"2018-01-12 01:36:47.000000000","message":"option 2 above would be to change this to 7","commit_id":"eea45a153aa8ac959bb4deae4ba3596ef62add06"},{"author":{"_account_id":7233,"name":"Matthew Oliver","email":"matt@oliver.net.au","username":"mattoliverau"},"change_message_id":"e57d2ff4d16ecdb1cfd34dba26cd0343cd0f84ff","unresolved":false,"context_lines":[{"line_number":4505,"context_line":"            rb, r\u0027r\\d+z\\d+-[^/]*$\u0027, verbose\u003dTrue)[\u0027graph\u0027])"},{"line_number":4506,"context_line":"        counts \u003d {k.split(\u0027-\u0027)[1]: d[\u0027placed_parts\u0027]"},{"line_number":4507,"context_line":"                  for k, d in report.items()}"},{"line_number":4508,"context_line":"        self.assertAlmostPartCount(counts, expected, delta\u003d5)"},{"line_number":4509,"context_line":""},{"line_number":4510,"context_line":""},{"line_number":4511,"context_line":"if __name__ \u003d\u003d \u0027__main__\u0027:"}],"source_content_type":"text/x-python","patch_set":1,"id":"7f96bb07_1d6868f3","line":4508,"range":{"start_line":4508,"start_character":59,"end_line":4508,"end_character":60},"updated":"2018-01-12 01:36:47.000000000","message":"and 7 here too","commit_id":"eea45a153aa8ac959bb4deae4ba3596ef62add06"}]}