10 messages in com.mysql.lists.clusterndb table indexes get corrupted upon ...
FromSent OnAttachments
Juha Heinanen24 Sep 2005 13:20 
Stewart Smith25 Sep 2005 18:45 
Juha Heinanen25 Sep 2005 23:24 
Juha Heinanen25 Sep 2005 23:43 
Stewart Smith26 Sep 2005 00:12 
Juha Heinanen26 Sep 2005 02:15 
Stewart Smith26 Sep 2005 06:48 
Juha Heinanen26 Sep 2005 07:10 
Alex Davies28 Sep 2005 16:46 
Alex Davies29 Sep 2005 03:05 
Subject:ndb table indexes get corrupted upon node restart
From:Juha Heinanen (jh@tutpro.com)
Date:09/24/2005 01:20:19 PM
List:com.mysql.lists.cluster

we have noticed that sometimes when master ndbd node is restarted by ndb_mgm "<id> restart" command or when ndbd host is rebooted, ndb table indexes get corrupted, i.e., "select * from table" works ok, but "select * from table where attr='value'" returns empty set, although there exists matching rows.

this is with debian sarge mysql-server 4.1.11a with two ndbd node/mysql server hosts and third host running ndb_mgmd.

any ideas on why the tables get corrupted and how to fix the problem?

-- juha