MySQL Forums
Forum List  »  InnoDB

Re: Heavy data lost in InnoDB after a bad-shutdown
Posted by: Rick James
Date: December 25, 2013 04:57PM

> Also this "innodb-autoextend-increment" option was surely not changed (I'm reading about it now for the first time), so I think this is set to default.

Please create a bug at bugs.mysql.com, pointing out that there seems to be some internal internal confusion over "64" versus "64M" in that setting.

Hibernate and JDBC connector should be doing the "right thing". But, without confirmation, I have deduced that they are doing the "wrong thing".

Please do one of these to help prove (or disprove) my deduction:
* Create a trivial program that goes through Hibernate that displays the output of SHOW VARIABLES LIKE 'auto%'.
* Turn on the 'general log', run a little bit of Hibernate stuff, turn the log off, then provide us the log.

The general log should show any changes to autocommit and also show COMMITs.

> unbelivable a 6day transaction between many server restart

True. A transaction that is 'open' at the time the server stopped will be ROLLBACKed after the server starts up again. This is part of the normal cleanup that guarantees data integrity.

> and shut down in the evening
> 2013-12-22 20:38:54 380 [Note] InnoDB: Database was not shutdown normally!
> 2013-12-22 20:38:54 380 [Note] InnoDB: Starting crash recovery.

It was not a graceful shut down. Please do simply do a power off. Lots of services, including mysql, may be stressed by such.

Options: ReplyQuote


Subject
Views
Written By
Posted
Re: Heavy data lost in InnoDB after a bad-shutdown
1507
December 25, 2013 04:57PM


Sorry, you can't reply to this topic. It has been closed.

Content reproduced on this site is the property of the respective copyright holders. It is not reviewed in advance by Oracle and does not necessarily represent the opinion of Oracle or any other party.