List:Cluster« Previous MessageNext Message »
From:Mike Kirby Date:March 24 2010 2:30pm
Subject:cluster replication
View as plain text  
Hi,

 

I noticed this type of message flooding the mysql.log on all my SQL
nodes in the cluster:

 

100324 10:43:20 [ERROR] NDB Binlog: Writing row (606078/7,0,0/0) to
ndb_binlog_index: 121

 

I have read a couple of bug reports, and it seems this is harmless, but
am not clear on how to stop the message, if possible.   

 

My setup is Mysql cluster with 4 SQL nodes, 4 data nodes. All 4 SQL
nodes are configured with binary logging enabled, and row-based
replication.  The slave is a single non-cluster db.  All slave tables
are MyISAM.

 

My questions are:

1)      Is  it possible to stop this message above from flooding the
logs?

2)      Do I actually need to have binary logging enabled on all 4 SQL
nodes, or just the one that the slave is connected to?  

3)      I notice my ndb_binlog_index table is pretty huge.  70+ million
rows.   What is the safest way to truncate what I don't need anymore?
Probably shut down replication, truncate the table, re-sync all data to
slave, and restart replication?

 

Thanks for your help.

 

 


Thread
cluster replicationMike Kirby24 Mar
  • Re: cluster replicationStewart Smith25 Mar