Details
-
Type:
Bug
-
Status: Closed
-
Priority:
Major
-
Resolution: Fixed
-
Affects Version/s: None
-
Fix Version/s: None
-
Component/s: None
-
Labels:
Description
Mysql bug #43014 is still repeatable with Maria 5.2. On recovery of a table containing a lot of indexes on big fields, the following assertion occurs:
mysqld: ma_blockrec.c:6187: _ma_apply_redo_insert_row_head_or_tail: Assertion `0' failed.
#8 0x00897de8 in __assert_fail () from /lib/libc.so.6
#9 0x0855b590 in _ma_apply_redo_insert_row_head_or_tail (info=0xa02c818, lsn=4296390922, page_type=1, new_page=0 '\000', header=0xa003ee2 "\003",
data=0xa003ee8 "", data_length=1968) at ma_blockrec.c:6187
#10 0x085a453e in exec_REDO_LOGREC_REDO_INSERT_ROW_HEAD (rec=0xbfabfa60) at ma_recovery.c:1373
#11 0x085a26db in display_and_apply_record (log_desc=0x9261528, rec=0xbfabfa60) at ma_recovery.c:528
#12 0x085a6fe3 in run_redo_phase (lsn=4294975495, apply=MARIA_LOG_APPLY) at ma_recovery.c:2518
#13 0x085a2047 in maria_apply_log (from_lsn=4294975495, apply=MARIA_LOG_APPLY, trace_file=0x9fbf7a0, should_run_undo_phase=1 '\001', skip_DDLs_arg=1 '\001',
take_checkpoints=1 '\001', warnings_count=0xbfac25e4) at ma_recovery.c:317
#14 0x085a1d28 in maria_recovery_from_log () at ma_recovery.c:218
#15 0x08527735 in ha_maria_init (p=0x9fb13a0) at ha_maria.cc:3251
#16 0x0839ed0d in ha_initialize_handlerton (plugin=0x9fa3090) at handler.cc:430
#17 0x0845b23b in plugin_initialize (plugin=0x9fa3090) at sql_plugin.cc:1239
#18 0x0845bab1 in plugin_init (argc=0x8a4a6d8, argv=0x9f71590, flags=0) at sql_plugin.cc:1467
#19 0x0826eb4b in init_server_components () at mysqld.cc:4126
#20 0x0826f611 in main (argc=16, argv=0xbfac2be4) at mysqld.cc:4600
Datadir from before recovery:
http://bugs.mysql.com/file.php?id=11479
RQG command line:
- 2010-07-07T02:01:39 # runall.pl \
- 2010-07-07T02:01:39 # --basedir=/home/philips/bzr/5.2/ \
- 2010-07-07T02:01:39 # --grammar=conf/engines/many_indexes.yy \
- 2010-07-07T02:01:39 # --gendata=conf/engines/many_indexes.zz \
- 2010-07-07T02:01:39 # --reporter=Recovery \
- 2010-07-07T02:01:39 # --duration=30 \
- 2010-07-07T02:01:39 # --queries=1M \
- 2010-07-07T02:01:39 # --engine=Maria \
- 2010-07-07T02:01:39 # --rows=1000 \
- 2010-07-07T02:01:39 # -
mysqld=-maria-checkpoint-interval=0 \ - 2010-07-07T02:01:39 # -
mysqld=-log-output=file \ - 2010-07-07T02:01:39 # --vardir=/home/philips/tmp/z2 \
- 2010-07-07T02:01:39 # -
mysqld=-loose-skip-innnodb \ - 2010-07-07T02:01:39 # --seed=time
Gliffy Diagrams
Attachments
Activity
- All
- Comments
- Work Log
- History
- Activity
- Transitions
A fresh data directory plus core file for the recovery crash.
bzr version-info:
revision-id: <email address hidden>
date: 2010-07-17 12:58:08 -0700
build-date: 2010-07-19 21:27:54 +0300
revno: 2826
branch-nick: 5.2
var-bug602604.zip
LPexportBug602604_var-bug602604.zip
Re: RQG: ma_blockrec.c:6187: _ma_apply_redo_insert_row_head_or_tail: Assertion `0' failed on Maria engine recovery
A fresh data directory plus core file for the recovery crash.
bzr version-info:
revision-id: igor@askmonty.org-20100717195808-mvh782jvt6c32u2d
date: 2010-07-17 12:58:08 -0700
build-date: 2010-07-19 21:27:54 +0300
revno: 2826
branch-nick: 5.2
trace + core + binary + vardirs
LPexportBug602604_var-bug602604trace.zip
Re: RQG: ma_blockrec.c:6187: _ma_apply_redo_insert_row_head_or_tail: Assertion `0' failed on Maria engine recovery
Launchpad bug id: 602604
Re: RQG: ma_blockrec.c:6187: _ma_apply_redo_insert_row_head_or_tail: Assertion `0' failed on Maria engine recovery
I tried to use the provided test files, but this didn't work for me.
The data in the .frm file is using a not supported character set::
How to verify this:
copy table100_maria_int_autoinc.* to datadir/test and do in the MySQL monitor:
show create table table100_maria_int_autoinc
I get the error:
Unknown collation '#60' in table 'table100_maria_int_autoinc' definition
I got this even when I recompiled MariaDB with --extra-charsets=all
When looking a the source, it looks like there is no character set 60 anymore, so I don't know how the test tables where created.
Will now try to see if I can run the test case instead.