MySQL Bugs: #37987: shutdown hangs after server was started with innodb_force

Bug #37987	shutdown hangs after server was started with innodb_force_recovery set
Submitted:	9 Jul 2008 11:30	Modified:	28 Apr 2009 17:56
Reporter:	Shane Bester (Platinum Quality Contributor)	Email Updates:
Status:	Duplicate	Impact on me:	None
Category:	MySQL Server: InnoDB storage engine	Severity:	S3 (Non-critical)
Version:	5.0.58	OS:	Any
Assigned to:		CPU Architecture:	Any
Tags:	innodb_force_recovery

Description:
after killing mysqld to recover from a runaway rollback and starting with innodb_force_recovery=3 , then dropping the table, mysqld hangs during the next shutdown.  another kill -9 is needed to shutdown the process.

How to repeat:
create table t1(id int)engine=innodb;
<kill mysqld>
mysqld --innodb_force_recovery=3
drop table t1;
mysqladmin shutdown

i'll try get a stack trace of the hanging thread.

I can not repeat the behaviour described with MySQL 5.0.62 on Windows, for example. Server was shutted down without any problems:

080709 20:53:36  InnoDB: Started; log sequence number 1 2827972647
InnoDB: !!! innodb_force_recovery is set to 3 !!!
080709 20:53:36 [Warning] Neither --relay-log nor --relay-log-index were used; s
o replication may break when this MySQL server acts as a slave and has his hostn
ame changed!! Please use '--relay-log=toshiba-user-relay-bin' to avoid this prob
lem.
080709 20:53:36 [Note] C:\Program Files\MySQL\MySQL Server 5.0\bin\mysqld-nt: re
ady for connections.
Version: '5.0.62-enterprise-gpl-nt'  socket: ''  port: 3308  MySQL Enterprise Se
rver (GPL)
080709 20:54:14 [Note] C:\Program Files\MySQL\MySQL Server 5.0\bin\mysqld-nt: No
rmal shutdown

080709 20:54:14  InnoDB: Starting shutdown...
080709 20:54:15  InnoDB: Shutdown completed; log sequence number 1 2827973254
080709 20:54:15 [Note] C:\Program Files\MySQL\MySQL Server 5.0\bin\mysqld-nt: Sh
utdown complete

my how to repeat was too simple. you really have to have a large rollback happening, then drop the table. i'll try again to repeat it.  If I remember correctly it was looping in logs_empty_and_mark_files_at_shutdown()

this is tentatively a duplicate of bug #44176