lp:~tsarev/percona-server/18205_02_wl36.patch

Branch merges

Propose for merging

No branches dependent on this one.

Merged into lp:percona-server/rnt-5.1 at revision 210

Laurynas Biveinis (community): Approve on 2012-02-01

Diff: 424 lines (+174/-48)

3 files modified

patches/series (+1/-1)
patches/wl36.patch (+170/-43)
patches/wl47.patch (+3/-4)

api

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

211. By Oleg Tsarev on 2012-02-06

Move changes in mysqlbinlog.result related to mysqlbinlog_fix.patch from mysql-test-rnt.diff to mysqlbinlog_fix.patch.

210. By Oleg Tsarev on 2012-01-31

Renamed patch mwl36.patch to wl36.patch.
Fixed code policy.
Added test.

209. By Oleg Tsarev on 2012-01-31

fixed code policy in mysqlbinlog_fix.patch

208. By Oleg Tsarev on 2012-01-16

Return back Percona's "innodb_stats_method=nulls_ignored" algorithm.

Related Launchpad bug #892405
Related MySQL bug http://bugs.mysql.com/bug.php?id=63320

I added new value "nulls_ignored_exact" for variable innodb_stats_method.

Current upsteam "nulls_ignored" algorithm calculate not_null values on every estimated (random) page, and after that try to approximates total value count.
Unfortunately, this algorithm works incorrectly on mostly null indexes (see MySQL Bug #63320).

Percona'a algorithm excludes from estimation pages with nulls, calculates the count of different values and approximates total value count.
This algorithm works correctly on mostly null indexes, but has limitation - it works correctly just on one-column-key indexes. On compound-keys algorithm can works incorrectly (I didn't test this).
Unfortunately, I can't imagine now how to discard this limitation.

207. By Oleg Tsarev on 2012-01-16

Fix incorrect month output from Subunit.pm
Perl localtime return month in range 0..11, subunit2junitxml expect range 1..12

206. By Oleg Tsarev on 2011-11-22

merge porting to 5.1.59

205. By Oleg Tsarev on 2011-10-19

Merge Alexey's commit:
    Bug #872754: Annotate rows events on slave doesnt' work without slave's binlog

    The problem was that Read_Master_Log_Pos was not calculated correctly
    when events were being skipped on the master (i.e. Annotate_Rows_Events
    when either replicate-annotate-rows-events or log-slave-updates were OFF).

    Fixed by changing the calculation so that master_log_pos is read from
    the event's header if it's available. Otherwise, proceed with the
    default logic.

204. By Oleg Tsarev on 2011-10-03

merge lp:percona-server/5.1

203. By Alexey Kopytov on 2011-09-24

Automerge from 5.1.

202. By Alexey Kopytov on 2011-09-23

Automerge from 5.1.