List:Cluster« Previous MessageNext Message »
From:artfire-kyle Date:November 19 2010 4:20pm
Subject:RE: MySQL CLuster Data Nodes Lose Connection
View as plain text  
First thank you for your response. 

I apologize as the title is a little misleading. It is the API (MySQLD)
nodes that are losing connection.
I have 4 servers total in the cluster. 2 x Management nodes and then the
data and mysqld run on the other two.

So yes the connection between mysql console and mysqld are what loses the
connection (i use "mysql -u root -p -h x.x.x.214" to connect to console),
but only from remote servers. If i access mysqld on API node 1 via API node
2 they establish fine. It is only servers from outside the cluster who try
to access mysqld that lose the connection (which are the servers that need
to connect via PHP).

I do have a firewall which is outside the cluster (and the rest of my web
servers) and is the main entry point for all traffic. I have all four of the
servers open on all ports (just until i can isolate this issue), and
everything is done with NAT.

When the remote connection is lost, I do not see any new entries in either
the management log file or the mysqld logging.

Here is the last entries in the mysqld log from one of the mysqld nodes.

090812 22:26:52 [Note] Event Scheduler: Loaded 0 events
090812 22:26:53 [Note] NDB Binlog: DISCOVER TABLE Event:
REPL$mysql/ndb_schema
090812 22:26:53 [Note] NDB Binlog: logging ./mysql/ndb_schema
(UPDATED,USE_WRITE)
090812 22:26:53 [Note] NDB Binlog: DISCOVER TABLE Event:
REPL$mysql/ndb_apply_status
090812 22:26:53 [Note] NDB Binlog: logging ./mysql/ndb_apply_status
(UPDATED,USE_WRITE)
2009-08-12 22:26:53 [NdbApi] INFO     -- Flushing incomplete GCI:s < 33737/6
2009-08-12 22:26:53 [NdbApi] INFO     -- Flushing incomplete GCI:s < 33737/6
090812 22:26:53 [Note] NDB Binlog: starting log at epoch 33737/6
090812 22:26:53 [Note] NDB Binlog: ndb tables writable
090812 22:26:53 [Note] mysqld: ready for connections.
Version: '5.1.39-ndb-7.0.9-1ubuntu7'  socket: '/var/run/mysqld/mysqld.sock' 
port: 3306  (Ubuntu)
090812 22:27:08 [Note] NDB Binlog: Node: 3, subscribe from node 5,
Subscriber bitmask 020
090812 22:27:08 [Note] NDB Binlog: Node: 4, subscribe from node 5,
Subscriber bitmask 020


I'm not sure how to procede, everything works great until the remote access
is lost to mysqld.

Thanks again.
-- 
View this message in context:
http://mysql-cluster.1050023.n4.nabble.com/MySQL-CLuster-Data-Nodes-Lose-Connection-tp3049544p3050611.html
Sent from the MySQL - Cluster mailing list archive at Nabble.com.
Thread
MySQL CLuster Data Nodes Lose Connectionartfire-kyle18 Nov
  • RE: MySQL CLuster Data Nodes Lose ConnectionAndrew Morgan19 Nov
    • RE: MySQL CLuster Data Nodes Lose Connectionartfire-kyle19 Nov
      • RE: MySQL CLuster Data Nodes Lose ConnectionAndrew Morgan19 Nov
      • RE: MySQL CLuster Data Nodes Lose ConnectionJeff Sturm19 Nov
        • RE: MySQL CLuster Data Nodes Lose Connectionartfire-kyle22 Nov
          • RE: MySQL CLuster Data Nodes Lose ConnectionJeff Sturm22 Nov
            • RE: MySQL CLuster Data Nodes Lose ConnectionBradley Watkins22 Nov
          • RE: MySQL CLuster Data Nodes Lose ConnectionMatthew Boehm22 Nov