mysqld: ma_blockrec.c:2998: write_block_record: Assertion `cur_block[1].page_count == 0' failed with aria-block-size=2K

[Philip Stoev]

When loading a table with large indexes and field sizes under maria-block-size=2K, the server asserted as follows:

mysqld: ma_blockrec.c:2998: write_block_record: Assertion `cur_block[1].page_count == 0' failed.

#8 0x006b0d98 in __assert_fail () from /lib/libc.so.6
#9 0x084f2cd6 in write_block_record (info=0x915b6800, old_record=0x0,
    record=0x915a6740 "\360\034\003rialgqmkmdcesfgqjgotgothcjkwsudrpknsxjxjxccvjwarmwlaaskwluooxhijnihuxhfmvzqnkplnldmpawbzosnvunqqpushgphfkedcxyiiinlgnrosdekjylyjrrfnbwcvviybutqoyrqiuvfrrgupmlvqndhksyzbvoklpmukeznigcurcbiwvjopjtmjq"..., row=0x915b684c, bitmap_blocks=0x915b684c, head_block_is_read=0 '\000',
    row_pos=0xa74b6410, undo_lsn=1, old_record_checksum=0) at ma_blockrec.c:2998
#10 0x084f416f in allocate_and_write_block_record (info=0x915b6800,
    record=0x915a6740 "\360\034\003rialgqmkmdcesfgqjgotgothcjkwsudrpknsxjxjxccvjwarmwlaaskwluooxhijnihuxhfmvzqnkplnldmpawbzosnvunqqpushgphfkedcxyiiinlgnrosdekjylyjrrfnbwcvviybutqoyrqiuvfrrgupmlvqndhksyzbvoklpmukeznigcurcbiwvjopjtmjq"..., row=0x915b684c, undo_lsn=1) at ma_blockrec.c:3515
#11 0x084f42b3 in _ma_write_init_block_record (info=0x915b6800,
    record=0x915a6740 "\360\034\003rialgqmkmdcesfgqjgotgothcjkwsudrpknsxjxjxccvjwarmwlaaskwluooxhijnihuxhfmvzqnkplnldmpawbzosnvunqqpushgphfkedcxyiiinlgnrosdekjylyjrrfnbwcvviybutqoyrqiuvfrrgupmlvqndhksyzbvoklpmukeznigcurcbiwvjopjtmjq"...) at ma_blockrec.c:3555
#12 0x0851a6e4 in maria_write (info=0x915b6800,
    record=0x915a6740 "\360\034\003rialgqmkmdcesfgqjgotgothcjkwsudrpknsxjxjxccvjwarmwlaaskwluooxhijnihuxhfmvzqnkplnldmpawbzosnvunqqpushgphfkedcxyiiinlgnrosdekjylyjrrfnbwcvviybutqoyrqiuvfrrgupmlvqndhksyzbvoklpmukeznigcurcbiwvjopjtmjq"...) at ma_write.c:146
#13 0x084ddaf9 in ha_maria::write_row (this=0xa73da168,
    buf=0x915a6740 "\360\034\003rialgqmkmdcesfgqjgotgothcjkwsudrpknsxjxjxccvjwarmwlaaskwluooxhijnihuxhfmvzqnkplnldmpawbzosnvunqqpushgphfkedcxyiiinlgnrosdekjylyjrrfnbwcvviybutqoyrqiuvfrrgupmlvqndhksyzbvoklpmukeznigcurcbiwvjopjtmjq"...) at ha_maria.cc:1040
#14 0x08396556 in handler::ha_write_row (this=0xa73da168,
    buf=0x915a6740 "\360\034\003rialgqmkmdcesfgqjgotgothcjkwsudrpknsxjxjxccvjwarmwlaaskwluooxhijnihuxhfmvzqnkplnldmpawbzosnvunqqpushgphfkedcxyiiinlgnrosdekjylyjrrfnbwcvviybutqoyrqiuvfrrgupmlvqndhksyzbvoklpmukeznigcurcbiwvjopjtmjq"...) at handler.cc:4708
#15 0x0830c2f0 in write_record (thd=0xaa41c50, table=0xa7397da0, info=0xa74b6758) at sql_insert.cc:1630
#16 0x0830a5ae in mysql_insert (thd=0xaa41c50, table_list=0xa7372d70, fields=..., values_list=..., update_fields=..., update_values=..., duplic=DUP_ERROR,
    ignore=true) at sql_insert.cc:860
#17 0x0827e5f7 in mysql_execute_command (thd=0xaa41c50) at sql_parse.cc:3275
#18 0x08287074 in mysql_parse (thd=0xaa41c50,
    rawbuf=0x91533ac8 "INSERT /*! IGNORE */ INTO table1000_maria_varchar_378_not_null VALUES ('sudrpknsxjxjxccvjwarmwlaaskwluooxhijnihuxhfmvzqnkplnldmpawbzosnvunqqpushgphfkedcxyiiinlgnrosdekjylyjrrfnbwcvviybutqoyrqiuvfrrgu"..., length=137415, found_semicolon=0xa74b721c) at sql_parse.cc:6156
#19 0x082797b3 in dispatch_command (command=COM_QUERY, thd=0xaa41c50,
    packet=0x91511a91 "INSERT /*! IGNORE */ INTO table1000_maria_varchar_378_not_null VALUES ('sudrpknsxjxjxccvjwarmwlaaskwluooxhijnihuxhfmvzqnkplnldmpawbzosnvunqqpushgphfkedcxyiiinlgnrosdekjylyjrrfnbwcvviybutqoyrqiuvfrrgu"..., packet_length=137416) at sql_parse.cc:1273
#20 0x08278a22 in do_command (thd=0xaa41c50) at sql_parse.cc:899
#21 0x08275b31 in handle_one_connection (arg=0xaa41c50) at sql_connect.cc:1138
#22 0x00821919 in start_thread () from /lib/libpthread.so.0
#23 0x0076acce in clone () from /lib/libc.so.6

[Philip Stoev (pstoev-askmonty)]

Unsimplified test case . Please run with --maria-block-size=2K

[Michael Widenius (monty)]

I have tested this with MariaDB 5.1 and 5.2 and both worked for me.

[Philip Stoev (pstoev-askmonty)]

The test case from this bug is still reproducible, but now with maria_block_size=1K

 perl mysql-test-run.pl --record --no-check-testcases t/bug726374.test --mysqld=--maria_block_size=1K

 bzr version-info
revision-id: <email address hidden>
date: 2011-07-21 15:21:22 +0300
build-date: 2011-07-23 13:14:58 +0300
revno: 3004
branch-nick: maria-5.2

[Philip Stoev (pstoev-askmonty)]

See also bug 815022

[Michael Widenius (monty)]

Fixed in 5.2 (the fix for bug 815022 also fixed this issue)