Description:
Hi,
when I cleanly restart MySQL with log_error_verbosity = 3, I have crash recovery messages in the error log. But after a clean stop, I do not expect crash recovery messages in the logs. See How to repeat for details.
How to repeat for 8.0.39, but I am also able to reproduce the same with 8.4.2 and 9.0.1.
Many thanks for looking into this,
Jean-François Gagné
How to repeat:
dbdeployer deploy single 8.0.39
function my_test() {
local pid=$(cat data/mysql_sandbox*.pid)
date
./use -N <<< "
set persist log_error_verbosity = 3;
set global innodb_fast_shutdown = 0;
shutdown"
while sleep 1; do ps -p $pid | grep -q $pid || break; done
date
./start > /dev/null
grep -i -e crash -e recovery data/msandbox.err
}
my_test
Wed 21 Aug 2024 15:10:16 EDT
Wed 21 Aug 2024 15:10:17 EDT
2024-08-21T19:10:17.671148Z 1 [Note] [MY-011980] [InnoDB] GTID recovery trx_no: 1329
2024-08-21T19:10:17.715277Z 0 [Note] [MY-010856] [Server] Failed to open the crashed binlog file when source server is recovering it.
2024-08-21T19:10:17.717157Z 0 [Note] [MY-013911] [Server] Crash recovery finished in binlog engine. No attempts to commit, rollback or prepare any transactions.
2024-08-21T19:10:17.717165Z 0 [Note] [MY-013911] [Server] Crash recovery finished in InnoDB engine. No attempts to commit, rollback or prepare any transactions.
2024-08-21T19:10:17.717529Z 0 [Note] [MY-012487] [InnoDB] DDL log recovery : begin
2024-08-21T19:10:17.717544Z 0 [Note] [MY-012488] [InnoDB] DDL log recovery : end
Suggested fix:
Do not show crash recovery messages after a clean shutdown.
And if some of these messages are meaningful, avoid terminology that can confuse the operator, including recovery, crashed, and crash recovery.
Description: Hi, when I cleanly restart MySQL with log_error_verbosity = 3, I have crash recovery messages in the error log. But after a clean stop, I do not expect crash recovery messages in the logs. See How to repeat for details. How to repeat for 8.0.39, but I am also able to reproduce the same with 8.4.2 and 9.0.1. Many thanks for looking into this, Jean-François Gagné How to repeat: dbdeployer deploy single 8.0.39 function my_test() { local pid=$(cat data/mysql_sandbox*.pid) date ./use -N <<< " set persist log_error_verbosity = 3; set global innodb_fast_shutdown = 0; shutdown" while sleep 1; do ps -p $pid | grep -q $pid || break; done date ./start > /dev/null grep -i -e crash -e recovery data/msandbox.err } my_test Wed 21 Aug 2024 15:10:16 EDT Wed 21 Aug 2024 15:10:17 EDT 2024-08-21T19:10:17.671148Z 1 [Note] [MY-011980] [InnoDB] GTID recovery trx_no: 1329 2024-08-21T19:10:17.715277Z 0 [Note] [MY-010856] [Server] Failed to open the crashed binlog file when source server is recovering it. 2024-08-21T19:10:17.717157Z 0 [Note] [MY-013911] [Server] Crash recovery finished in binlog engine. No attempts to commit, rollback or prepare any transactions. 2024-08-21T19:10:17.717165Z 0 [Note] [MY-013911] [Server] Crash recovery finished in InnoDB engine. No attempts to commit, rollback or prepare any transactions. 2024-08-21T19:10:17.717529Z 0 [Note] [MY-012487] [InnoDB] DDL log recovery : begin 2024-08-21T19:10:17.717544Z 0 [Note] [MY-012488] [InnoDB] DDL log recovery : end Suggested fix: Do not show crash recovery messages after a clean shutdown. And if some of these messages are meaningful, avoid terminology that can confuse the operator, including recovery, crashed, and crash recovery.