MySQLd servers initialization

The two MySQLd servers need to have their DB initialized (they need to know at least who users are and who can connect from). For this purpose we use the install_db script:

# $NODE1_HOME/scripts/mysql_install_db –force –datadir=$NODE1_DATADIR –basedir=$NODE1_HOME
$NODE2_HOME/scripts/mysql_install_db –force –datadir=$NODE2_DATADIR –basedir=$NODE2_HOME

Starting nodes
It is now the time to start the management node:

# $MGMT_BIN/ndb_mgmd -f $MGMT_ETC/config.ini #The mgmt should start
# less $MGMT_VAR/lib/mysql-cluster/ndb_1_cluster.log
#Check the log for errors

Now we need to start the data nodes:

# cd $NODE1_NDBD_DATADIR
# $NODE1_BIN/ndbd –initial -c [MGMT_IP_ADDRESS] –bind-address=[FIRST_NODE_IP_ADDRESS]
# cd $NODE2_NDBD_DATADIR
# $NODE2_BIN/ndbd –initial -c [MGMT_IP_ADDRESS] –bind-address=[SECOND_NODE_IP_ADDRESS]

Please note the –initial parameter used as argument to the ndbd command: this is needed only for the very first run… omit it next time!

Now the first check: we ask the MGMT console the cluster status:

# $MGMT_BIN/ndb_mgm -c [MGMT_IP_ADDRESS] -e show
Connected to Management Server at: 10.145.2.3:1186
Cluster Configuration
———————
[ndbd(NDB)] 2 node(s)
id=2 @10.145.2.33 (mysql-5.1.23 ndb-6.2.15, Nodegroup: 0, Master)
id=3 @10.145.2.34 (mysql-5.1.23 ndb-6.2.15, Nodegroup: 0)
[ndb_mgmd(MGM)] 1 node(s)
id=1 @10.145.2.3 (mysql-5.1.23 ndb-6.2.15)
[mysqld(API)] 2 node(s)
id=4 (not connected, accepting connect from any host)
id=5 (not connected, accepting connect from any host)

You should see something similar with your IPs: note that we actually see the two datanodes just started and the mgmt node.
At the end we need to start the two MySQLd that acts as frontend:

# $NODE1_BIN/mysqld –defaults-file=$NODE1_/my.cnf.master &
# $NODE2_BIN/mysqld –defaults-file=$NODE2_ETC/my.cnf &

and executing the query of the mgmt luster console we should now see all nodes:

# $MGMT_BIN/ndb_mgm -c [MGMT_IP_ADDRESS] -e show
Connected to Management Server at: 10.145.2.3:1186
Cluster Configuration
———————
[ndbd(NDB)] 2 node(s)
id=2 @10.145.2.33 (mysql-5.1.23 ndb-6.2.15, Nodegroup: 0, Master)
id=3 @10.145.2.34 (mysql-5.1.23 ndb-6.2.15, Nodegroup: 0)
[ndb_mgmd(MGM)] 1 node(s)
id=1 @10.145.2.3 (mysql-5.1.23 ndb-6.2.15)
[mysqld(API)] 2 node(s)
id=4 @10.145.2.3 (mysql-5.1.23 ndb-6.2.15)
id=5 @10.145.2.3 (mysql-5.1.23 ndb-6.2.15)

If you see something different then probably you have a problem: please look at log files that you can find in $NODE1_VAR/log or $NODE2_VAR/log. If you provide as much information as possible I’ll try to help you out.

button1 addthis How to install a MySQL cluster on a single UNIX/Linux server     AddThis Feed Button

Pages: 1 2 3 4 5 6 7

14 Responses to “How to install a MySQL cluster on a single UNIX/Linux server”

  1. Karuna says:

    Hi,
    Thanks a lot. This demonstration is very useful to me and it is working fine. I have installed and tested the mysql clusters in my single linux machine.

    Very useful……………….. :)

  2. apiero says:

    Hi Karuna,

    many thanks for your feedback: I tried all these steps for a clean installation but I’m also the writer of this tutorial ;)

    Bye
    Apiero

  3. Snehalkumar Patel says:

    Hi..
    I am facing problem after completing cluster configuration, cluster runs properly but i dont see “Master” in [ndbd(NDB)] 2 nodes, both nodes show:
    [ndbd(NDB)] 2 node(s)
    id=2 @192.168.42.174 (mysql-5.1.23 ndb-6.2.15, starting, Nodegroup: 0)
    id=3 @192.168.42.175 (mysql-5.1.23 ndb-6.2.15, starting, Nodegroup: 0)

    :-(
    I am using Redhat Enterprise 5.
    Please Help

  4. apiero says:

    Hi,

    did you verify that the firewall was disabled? Maybe it is blocking some ports
    Try, as root, with:
    /etc/init.d/iptables stop
    /etc/init.d/ip6tables stop

    Please let me know if it works or not.

    Bye
    Apiero

  5. Kagan says:

    when trying to run the deamon in the final step get the following error msg:

    /usr/lib/mysql_cluster/node_1/bin/mysqld: Table ‘mysql.plugin’ doesn’t exist
    090122 18:01:18 [ERROR] Can’t open the mysql.plugin table. Please run mysql_upgrade to create it.
    090122 18:01:18 InnoDB: Started; log sequence number 0 46409
    090122 18:01:18 [Note] Starting Cluster Binlog Thread
    090122 18:01:18 [ERROR] Fatal error: Can’t open and lock privilege tables: Table ‘mysql.host’ doesn’t exist

    any idea why?

  6. apiero says:

    Hi Kagan,

    did you use the exact versions of MySQL that I did? It looks like you missed the first part of page 6 (http://www.brucalipto.org/linux/how-to-install-a-mysql-cluster/6).

    Please let me know if it works or not.

    Bye
    Apiero

  7. Alexwebmaster says:

    Hello webmaster
    I would like to share with you a link to your site
    write me here preonrelt@mail.ru

  8. Fernando says:

    hi mate, i’ve managed to get almost everything working..
    instead of doing everything on one host, i did it with 3 VMs, one is management, and 2 nodes/storage.
    I got the nodes to connect without problems. but i can’t get the mysql database to connect. any ideas?

    Cluster Configuration
    ———————
    [ndbd(NDB)] 2 node(s)
    id=2 @172.12.13.105 (Version: 7.0.6, starting, Nodegroup: 0)
    id=3 @172.12.13.106 (Version: 7.0.6, starting, Nodegroup: 0)

    [ndb_mgmd(MGM)] 1 node(s)
    id=1 @mgm01 (Version: 7.0.6)

    [mysqld(API)] 2 node(s)
    id=4 (not connected, accepting connect from any host)
    id=5 (not connected, accepting connect from any host)

    thanks

  9. totojack says:

    Hi.

    Do you think is it better a cluster or a replication (1 master – n slave) configuration in a situation of an high load website?
    And have you ever tried mysql proxy as loadbalancer?

  10. apiero says:

    Hi,

    I think that in a situation of an high load website a cluster would be better but keep in mind that the MySQL cluster has many “limitation”: not really limitations but its own rules. Keep in mind that MySQL is very fast so maybe a simple node could be sufficient.
    I tried to look at MySQL proxy as loadbalancer but I did not understand how does it work: actually if I would need to setup something “big” with MySQL I would try MMM (Multi-Master Replication Manager) for MySQL[1].

    Please let me know what was your choice!
    Bye
    Apiero
    [1]http://mysql-mmm.org/

  11. Rebel says:

    Hi,

    (still alive ?)

    im having a question …

    i am running the ndb managment server and the mysql server one on the same server and the mysql server 2 on a second server.

    if the server 1 (with mysql1 and mgmt server) crashes > there is data loss > what will append when i will restart the whole servers ?

    the cluster system is only transmitting the queries ? isn’t it?

  12. I normally don’t write-up in Blogs but your weblog forced me to, awesome operate.. stunning

  13. Been looking for this article for long time ago and finally found here. thanks for sharing this post. appreciate!

Leave a Reply