ma_blockrec.c:6351: _ma_apply_redo_insert_row_head_or_tail: Assertion `0' failed during Aria recovery

[Philip Stoev]

maria_read_log produced the following error:

The transaction log starts from lsn (1,0x2007)
TRACE of the last maria_read_log
/home/buildbot/rqg-buildbot-slave/rqg-weekend/build/mysql-test//../storage/maria/maria_read_log: Maria engine: starting recovery
recovered pages: 0% 10% 20% 30%maria_read_log: ma_blockrec.c:6351: _ma_apply_redo_insert_row_head_or_tail: Assertion `0' failed.

rqg command line:

# 2010-09-07T16:09:37 # /home/buildbot/randgen/runall.pl \
# 2010-09-07T16:09:37 # --queries=1M \
# 2010-09-07T16:09:37 # --engine=Maria \
# 2010-09-07T16:09:37 # --mysqld=--log-output=file \
# 2010-09-07T16:09:37 # --reporters=ErrorLog,Backtrace,Recovery,Shutdown \
# 2010-09-07T16:09:37 # --duration=240 \
# 2010-09-07T16:09:37 # --threads=10 \
# 2010-09-07T16:09:37 # --rows=10 \
# 2010-09-07T16:09:37 # --mask-level=1 \
# 2010-09-07T16:09:37 # --mysqld=--loose-maria-group-commit=soft \
# 2010-09-07T16:09:37 # --mysqld=--loose-maria_group_commit_interval=1 \
# 2010-09-07T16:09:37 # --mysqld=--maria-checkpoint-interval=0 \
# 2010-09-07T16:09:37 # --mysqld=--maria-block-size=4K \
# 2010-09-07T16:09:37 # --grammar=conf/engines/varchar.yy \
# 2010-09-07T16:09:37 # --gendata=conf/engines/varchar.zz \
# 2010-09-07T16:09:37 # --mask=16255 \
# 2010-09-07T16:09:37 # --queries=100000000 \
# 2010-09-07T16:09:37 # --mask=16255 \
# 2010-09-07T16:09:37 # --basedir=. \
# 2010-09-07T16:09:37 # --seed=time \
# 2010-09-07T16:09:37 # --vardir=/home/buildbot/rqg-buildbot-slave/rqg-weekend/vardir-maria_recovery_weekend/current

bzr version-info:

revision-id: <email address hidden>
date: 2010-09-06 02:48:51 +0300
build-date: 2010-09-08 10:30:40 +0300
revno: 2922
branch-nick: 5.1

[Philip Stoev (pstoev-askmonty)]

Still repeatable with

revision-id: <email address hidden>
date: 2010-09-07 19:58:39 +0300
build-date: 2010-09-09 10:35:46 +0300
revno: 2924
branch-nick: 5.1

[Michael Widenius (monty)]

Did run the given test case for 12 hours without problems.
I also downloaded the test data and run maria_read_log -a -s on it without any probelms.
As I have fixed a lot of related things lately, I have to assume this bug is also fixed.