lp:~mdcallag/sysbench/0.4-dev

Branch merges

Propose for merging

No branches dependent on this one.

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

81. By Mark Callaghan on 2014-05-30

make counter work despite optimizer

80. By Mark Callaghan on 2014-04-15

use more precision for times
change update-non-key field to match what is used by jmongosysbench

79. By Mark Callaghan on 2012-03-12

add --oltp-inlist-sum that adds the query:
  select sum(k) from TABLE where id in (....)
and there are 80 non-consecutive id values in the inlist.

78. By Mark Callaghan on 2011-11-28

add incupdate2 test
make some stats per-interval

77. By Mark Callaghan on 2010-11-06

Fix a bug in the previous change to fix the fileio race

76. By Mark Callaghan on 2010-11-05

Fix race conditions in sysbench fileio. There were a few global variables updated without
mutex protection. Remove fsync_mutex as the global sysbench mutex is sufficient. I don't
think fine-grained locking is needed in this code. I was getting errors and early
test failures when running this with multiple threads and fsync enabled.

75. By Mark Callaghan on 2010-10-19

Use somewhat random data for the 'k' and 'c' and 'pad' columns to make it harder to compress.

74. By Mark Callaghan on 2010-09-29

Change --batch stats to be per interval, not cumulative

73. By Mark Callaghan on 2010-07-30

Add
--oltp-simple-update to get read-write workload with updates on non-indexed column for simple test mode.
--oltp-point-select-mysql-handler-read-open-close to use HANDLER_OPEN_READ_CLOSE for tests that use HANDLER

72. By Mark Callaghan on 2010-05-13

Add --oltp-test-mode=incupdate and --oltp-test-mode=incinsert for new tests.
With 'incupdate' each thread does nothing but updates to increment the integer
stored in column 'c'. With 'incinsert' the same is done using
INSERT ... ON DUPLICATE KEY UPDATE c = c + 1. The table of 'c' in the sbtest
table is not changed from char. I have a use case for which the performance
of these statements matter.