I've switched over to the mysqlmanager startup system instead of the old
mysqld_safe because thats the only supported method in mysql5.

I needed to restart a DB so I did a `/etc/init.d/mysqlmanager restart`
which seemed to work, but there were some problems:

- the daemon was no longer accepting connections
- ps showed 2 copies of mysqld running

I also noticed a lot of errors like this in mysqld.err:

070124 15:27:02 [ERROR] /usr/sbin/mysqld: Incorrect information in file:
'./databasename/table.frm'

One of the daemon processes would not respond to kill so eventually I
fired up gdb and killed it (it was stuck on a futex operation). I then
restarted mysql and it went through a huge crash recovery.

A co-worker tells me that mysqlmanager has been known to do this for
ages. Is this true? The DB in question is all InnoDB, approx 150GB in
12 tables. mysql version 5.0.30.

Thanks!