Going on with our MySQL Cluster studies, now I am starting this new post to talk or write about the Storage or Data Nodes, that one is responsible to storage all cluster data. When we start a cluster (as you have read at MySQL Cluster First Topics), the first step is to start the Management Node and the second part is to start the storage nodes, responsible to storage data using NDBCLUSTER Storage Engine. Normally, the correct, stated and desired form to have nodes to support storage node is to concept a separate machine in order to have only ndbd process using that machine resources. It is important cause, if data node or its daemon named ndbd do not have enough memory to maintain at least indexes on memory, it will crash and will not function properly. But, we will treat about this little things most ahead in order to introduce first steps to concept a data node, start and be happy with you cluster!

After to chose what machines or boxes will be the cluster storage nodes, paying attention about the configurations as the same model of CPU, same amount of memory and the most important part, concept a full 64-bit machine, including OS, hardware and softwares, we can start to download MySQL Cluster Storage Node software component in order to install correct packages and configure it to connect with Management node. Remember, all hardware involved must be the same configuration in order to avoid performance problems and keep the cluster simple as much as you can (normally, I have been implementing MySQL Cluster using Virtual Machines in order to have the max proximity of hardware configuration – the problem is, we must have a look on SPOFS, or, single point of failure). To build Storage Nodes, it will be required to download two packages:

• MySQL-Cluster-gpl-tools-7.x.x-x
• MySQL-Cluster-gpl-storage-7.x.x-x

As I am using CentOS 5.5 to write this post, I have downloaded “.rpm” packages that will be installed using rpm package manager at terminal linux level. You can apply this post on MS Windows, for example and install execs packages as you want. Below, I will demonstrate the install process:

[ root@node3 ]# rpm -ivh MySQL-*
Preparing... ############################################ [100%]
1:MySQL-Cluster-gpl-tools ################################## [100%]
2:MySQL-Cluster-gpl-stora ################################## [100%]
[ root@node1 ]#

After this, we can use the files cluster concepts, what is the local files and global files! Local files is that files that is created locally on the node’s machine and will serve to configure the cluster Storage Nodes connectstring (ndb_connectstring variable or its shortcut “-c” can be used on local files or by passing through command line). A good touch is, when you are using local files, you will able to inform just little things that will be applied on Storage Nodes connection with Management Nodes. As we have necessary components installed at this moment, we must create a configuration file that one will be read when ndbd starts (you can query where is the default location that ndbd will read local files using ndbd –help and on the command line and reading the firsts lines). The local file will be created below on the example:

[ root@node3 ]# vim /etc/my.cnf

# my.cnf local file
# storage node's configuration
#
[ndbd]
ndb-connectstring=192.168.1.100:1186
# finish

Before I forget, the cluster configuration global file is that just one we have created on the first post, that one the majority of the configurations were mentioned and resides on the Management Node. There, we can mention that configurations that will be applied on all Storage Nodes using the section [ndbd default].

Now, you can simply call ndbd on the command line and it will read cluster configuration local file in order to know the exact location of the Management Node (ndb_mgmd) and initiate your tests though the normal levels before appear as started on ndb_mgm Management client. Remember that the location where you can check about all nodes are running is the ndb_mgm client on Management Node (if you are using two Management Nodes – this is good, what a wonderful thing! – you can use both to retrieve all cluster status information).

Calling ndbd on the command line:

[ root@node3 ]# ndbd
2011-03-25 13:21:13 [ndbd] INFO -- Angel connected to '192.168.1.100:1186'
2011-03-25 13:21:13 [ndbd] INFO -- Angel allocated nodeid: 3

As you can see, after start the ndbd process, two processes were started together, one is the ndbd Storage Node and another is the ndbd Angel, the process tha will rise up main Storage Node process in case it going down. The started Storage Node received its NodeID as previously configured and now is waiting the other nodes to finish its complete start! All Storage Nodes envolved on the MySQL Cluster must pass through the same process – installation components, creation of configuration file mentioning the ndb_connectstring under [ndbd] section and start of ndbd. After to complete these jobs on all cluster storage nodes, go to the Management Node and query for the nodes status using ndb_mgm client as mentioned below:

[ root@node1 ~ ] # ndb_mgm -e "SHOW"
-- NDB Cluster -- Management Client --
Cluster Configuration
---------------------
[ndbd(NDB)] 2 node(s)
id=3 @192.168.1.103 (mysql-5.1.51 ndb-7.1.10, NodeGroup: 0, Master)
id=4 @192.168.1.104 (mysql-5.1.51 ndb-7.1.10, NodeGroup: 0)

[mysqld(API)] 4 node(s)
id=11 (not connected, accepting connect from 192.168.0.103)
id=12 (not connected, accepting connect from 192.168.0.104)

See ya.

You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.