origin/10.6 8ff10969996a5349e906fab44b45c6e3ec53eea0 2022-08-24T17:06:57+03:00

I have two concurrent sessions.

thread1:

    CREATE TABLE IF NOT EXISTS t3 ( col1 INT, col_text TEXT , col_text_g TEXT GENERATED ALWAYS AS (SUBSTR(col_text,1,499)) ) ENGINE = InnoDB ROW_FORMAT = Compact |

    ALTER TABLE t3 ADD KEY ( col_text_g(9) , col_text(9) ) ;

thread2:

    CHECK TABLE t3 |

    DELETE FROM t3 WHERE col1 = 8 OR col1 IS NULL ; INSERT INTO t3 (col1, col_text) VALUES (8,''), (8,'') ;

After some runtime

CHECK TABLE t3 harvests

   Warning InnoDB: Index 'col_text_g' contains 4 entries, should be 2.

and RQG reacts with killing the server.

pluto:/data/results/1661424131/TBR-36$ _RR_TRACE_DIR=./1/rr/ rr replay

RQG

====

# git clone https://github.com/mleich1/rqg --branch experimental RQG

#

# GIT_SHOW: HEAD -> experimental, origin/experimental 32e1b1a96b072f6a71c564d3e2d50d9cf2fb3561 2022-08-24T17:04:19+02:00

# rqg.pl  : Version 4.0.6 (2022-05)

#

# $RQG_HOME/rqg.pl \

# --duration=300 \

# --queries=10000000 \

# --no_mask \

# --seed=random \

# --rpl_mode=none \

# --gendata_sql=conf/mariadb/table_stress.sql \

# --max_gd_duration=1200 \

# --gendata=conf/mariadb/table_stress.zz \

# --engine=InnoDB \

# --mysqld=--loose-idle_write_transaction_timeout=0 \

# --mysqld=--wait_timeout=28800 \

# --mysqld=--loose-idle_transaction_timeout=0 \

# --mysqld=--log-output=none \

# --mysqld=--interactive_timeout=28800 \

# --mysqld=--sync-binlog=1 \

# --mysqld=--lock-wait-timeout=86400 \

# --mysqld=--loose-innodb_fatal_semaphore_wait_threshold=300 \

# --mysqld=--net_read_timeout=30 \

# --mysqld=--connect_timeout=60 \

# --mysqld=--plugin-load-add=file_key_management.so \

# --mysqld=--loose-table_lock_wait_timeout=50 \

# --mysqld=--innodb-buffer-pool-size=5M \

# --mysqld=--net_write_timeout=60 \

# --mysqld=--innodb-lock-wait-timeout=50 \

# --mysqld=--loose_innodb_lock_schedule_algorithm=fcfs \

# --mysqld=--log_bin_trust_function_creators=1 \

# --mysqld=--loose-max-statement-time=30 \

# --mysqld=--log-bin \

# --mysqld=--loose-plugin-load-add=provider_lz4.so \

# --mysqld=--file-key-management-filename=$RQG_HOME/conf/mariadb/encryption_keys.txt \

# --mysqld=--innodb_page_size=4K \

# --mysqld=--loose-idle_readonly_transaction_timeout=0 \

# --mysqld=--loose-innodb_read_only_compressed=OFF \

# --mysqld=--loose-debug_assert_on_not_freed_memory=0 \

# --mysqld=--slave_net_timeout=60 \

# --reporters=Backtrace,Deadlock1,ErrorLog \

# --validators=None \

# --grammar=TBR-36.yy \

# --threads=2 \

# --batch \

# --script_debug=_nix_ \

# --rr=Extended \

# --rr_options=--chaos --wait  \

# <local settings>

The already simplified tests replays also frequent the error patterns

-  [ 'TBR-915-MDEV-25004', 'mysqld: .{1,200}row0upd.cc:.{1,20} dberr_t row_upd_sec_index_entry\(upd_node_t\*, que_thr_t\*\): Assertion .0. failed' ],   

    but in the current case system versioning is not involved.

- [ 'TBR-915A', '\[ERROR\] InnoDB: Record in index .{1,70} of table .{1,150} was not found on update: TUPLE .+mysqld: .{1,200}row0upd.cc:.{1,20} dberr_t row_upd_sec_index_entry\(upd_node_t\*, que_thr_t\*\): Assertion .0. failed' ],

- [ 'TBR-835', '\[ERROR\] InnoDB: Record in index `(?!FTS_DOC_ID_INDEX).{0,200}` of table .{1,100} was not found on update: ' ],

I am aware that the patterns are to some extend redundant.