Configurator for MySQL Cluster - version 4.2.0

This page contains the following information:

What will be installed
System Requirements
How to Install the Deployment Package
Documentation of CoreScripts
File structure and organization
Support
Troubleshooting

What will be installed

If you chose to use binary/RPM version, you will be using MySQL Cluster ...
If you chose to build from source, you will be using MySQL Cluster .. (GPL with innodb).

Below are instruction how to get going! Print this page as a reference.

At the end you will also find some common problems and how to resolve them. Email feedback@severalnines.com if you have problems.

System Requirements

No prior installation of mysql (watch out for previously auto-installed mysql packages that comes in the OS. Use e.g 'yum remove mysql mysql-server' or 'apt-get remove mysql'. Please note that the package names may be called something different.

Ubuntu users (and possibly others) - enable root account on all machines.

SELinux and firewalls disabled
SELinux - Check /etc/selinux/config (set SELINUX=disabled and reboot
Firewall - /sbin/iptables -L - this command should print out that no ports are blocked

chkconfig iptables off 
/etc/init.d/iptables stop

/etc/hosts is sane. It must not have a line looking like this:
127.0.0.1 some_hostname localhost localhost.localdomain localhost
If you have some_hostname the nodes will not connect correctly, and yo have to change to (remove all references to actual hostname on the line(s) starting with 127.0.0.1:
127.0.0.1 localhost localhost.localdomain localhost

Otherwise nodes will not connect to each other - 'waiting for PID' or will always be stuck in start phase 0 - will be the result

You should also disable NUMA, and make sure you don't swap!

How to Install the Deployment Package

You should copy it to a front end node (ClusterControl™ Server) that has root access to the hosts (or is part) of your cluster.

You can chose to use a binary tar distribution of MySQL Cluster or to build from source .

Using a binary build (or RPMs) from MySQL

Run (in order):

If you don't have internet access

Please contact us for a list of necessary RPM/TARs to install.

To build from source

You must have installed

gcc and g++
apt-get install gcc g++ autoconf automake libtool or yum install gcc gcc-c++ autoconf automake libtool
wget
apt-get install wget or yum install wget
ncurses-dev
apt-get install libncurses5-dev or yum install ncurses-devel

You should run these scripts from mysqlcluster-63/cluster/scripts/install:

shared-ssh-keys.sh - Sets up shared ssh keys, follow the instructions carefully. This step is not neeed if you already have shared ssh keys (ssh without typing passwords).
download-and-compile.sh - gets the latest officially released cluster source, configure, and build. The resulting build is in the build/ directory
install-cluster.sh - Distribute the binary located in 'mysqlcluste-xy/cluster/repo' directory ... to the hosts in the cluster
bootstrap.sh - sets up directory structures and installs the mysql servers configured with a hostname. The script will delete any old information (if a previous mysqld data directory exists on the same location) located on a host. Configuration files (my.cnf and config.ini) will be installed in /
'cd ..' and then do start-cluster.sh --initial - starts the management server(s), the data nodes and the mysql server(s) configured with a hostname
Instead of the above four scripts you can run deploy.sh, which will run download-and-compile.sh, dist-src-build.sh,bootstrap.sh, and start-cluster.sh --initial

Documentation of CoreScripts

Installation scripts	Description	Version (when script was introduced)
location `mysqlcluster-xy/cluster/scripts/install`	The installation scripts are located in `mysqlcluster-xy/cluster/scripts/install`
deploy.sh	The script runs: shared-ssh-keys.sh/li> download-binary.sh or install-rpm.sh install-cluster.sh bootstrap.sh start-cluster.sh --initial	1.0
download-and-compile.sh	usage: `./download-and-compile.sh` Download and compiles the latest MySQL Cluster source (either 6.3 or 7.0 depending on what you have selected). Source releases are made more often than binary releases. The binary tarball from the build is placed in `mysqlcluster-xy/cluster/repo`	1.0
download-binary.sh	usage: `./download-binary.sh` Download and compiles the latest MySQL Cluster binary release (either 6.3 or 7.0 depending on what you have selected). Only works with Linux x86-64 Places the binary tarball in `mysqlcluster-xy/cluster/repo`.	2.6
install-cluster.sh	usage: `./install-cluster.sh` Distributes the mysql binary tarball in `mysqlcluster-xy/cluster/repo` to the defined hosts.	2.6
install-host.sh	usage: `./install-host.sh <hostname>` Distributes the mysql binary tarball in `mysqlcluster-xy/cluster/repo` to a named host.	2.6
bootstrap.sh	usage: `./bootstrap.sh` Run only the first time you setup a new Cluster! The bootstrap script does: removes mysql data directories on each host. removes ndbd data directories on each host. removes ndb_mgmd data directories on each host. creates mysql data directories on each hosti creates ndbd data directories on each hosti creates ndb_mgmd data directories on each hosti copy configuration files (config.ini and my.cnf) to relevant hosts create 'mysql' user on each host that is defined to run a mysqld This script must be executed before starting the cluster the first time.	1.0
add-mysqld.sh	usage: `./add-mysqld.sh --hostname=<hostname> --hostname=<hostname> - the hostname where to install the mysql server.` add-mysqld.sh will copy the binaries in 'repo' to the host, install the binary, create the mysql user, and datadir, and install the mysql server. start/stop scripts are also generated.	2.6
Cluster start/stop scripts	Description
location `mysqlcluster-xy/cluster/scripts/`	The following scripts are located in `mysqlcluster-xy/cluster/scripts/`
start-cluster.sh [--initial]	usage: `./start-cluster.sh [--initial]` Performs a start of mysql cluster. If `--initial` is specifed it will DELETE the filesystem for the data nodes (this will clear out all old data stored in this MySQL Cluster). The startup sequence is as follows: start all managment servers start all data nodes start all mysql servers	1.0
stop-cluster.sh	usage: `./stop-cluster.sh` Stops the entire cluster. The script will: stop all mysql servers stop all data nodes stop all management servers	1.0
Management server scripts	Description
location `mysqlcluster-xy/cluster/scripts/`	The following scripts are located in `mysqlcluster-xy/cluster/scripts/`
./start-mgmd.sh --hostname=<host>	usage: `./start-mgmd.sh --hostname=<host>` starts the management server on <host>	1.0
./stop-mgmd.sh --hostname=<host> stop-mgmd-<host>.sh	./stop-mgmd.sh --hostname=<host> stops the management server on <host>	1.0
Data node scripts	Description
start-ndbd.sh --hostname=<host> [--id=<id>] [ --initial ] [--nowait] [--force]	usage: `./start-ndbd.sh --hostname=<host> [--id=<id>] [ --initial ] [--nowait] [--force]` starts the data node with <id> on <host> with --initial. This means it will clear out its data directory, and perform an initial node recovery (copy all data from the other node in the same node group. The ID is used to be able to distinguish between one or more data nodes located on one host.	1.0
stop-ndbd.sh --hostname=<host> [--id=<id>]	usage: `./stop-ndbd.sh --hostname=<host> [--id=<id>]` stops the data node with <id> on <host>. The ID is used to be able to distinguish between one or more data nodes located on one host.	1.0
Mysql server scripts	Description
start-mysqld.sh --hostname=<host>	usage: `./start-mysqld.sh --hostname=<host>` starts the mysqld on <host>. If there is a cluster that is started, then the mysqld will join the cluster.	1.0
stop-mysqld.sh --hostname=<host>	usage: `stop-mysqld.sh --hostname=<host>` stops the mysqld on <host>	1.0
mysqlclient.sh --hostname=<host>	usage: `./mysqlclient.sh --hostname=<host>` starts a mysqlclient on the mysqld running on <host>	1.0
Administration scripts	Description
location `mysqlcluster-xy/cluster/scripts/`	The following scripts are located in `mysqlcluster-xy/cluster/scripts/`
rolling-restart.sh [--initial] --layer=[all\|storage\|sql]	usage: `./rolling-restart.sh [--initial]` Performs a rolling restart. Usually this is done when upgrading the config.ini (mysqlcluster-xx/config/config.ini). . The rolling restart is online, no data will be lost, and there is no service interruption. E.g, if you have increased the DataMemory (updated the mysqlcluster-xx/config/config.ini), then you need to perform a rolling restart. Some parameter changes requires an `rolling-restart.sh --initial`. See the reference manual for more information. The following will happen: copy mysqlcluster-xx/config/config.ini to the hosts running the managment servers restart the management servers (one at time) restart the data nodes (one at time) - the data nodes will now be started with the new config.ini directives	1.0
show.sh	usage: `./show.sh` Prints out the nodes in the cluster	1.0
status.sh	usage: `./status.sh` Prints out the status of the data nodes in the cluster	1.0
start-backup.sh	usage: `./start-backup.sh <backupdir>` Starts a backup, and aggregates the backup files in <backupdir>	1.0
restore-backup.sh	usage: `./restore-backup.sh --backupdir=<backupdir>` restores the backup files in <backupdir> By specifying the following you can generate csv files from the backup files in <backupdir>: `--csv` `--csvdir=<output dir for csv files (.txt)>` Nothing is restored on the data nodes when specifying --csv.	2.5
ndb_mgm.sh	Starts the management client.	1.0
Tool scripts	Description
location `mysqlcluster-xy/cluster/scripts/tools`	The following scripts are located in `mysqlcluster-xy/cluster/scripts/tools`
check-cluster-log.sh	usage: `./check-cluster.log.sh --hostname=<hostnamr>` Tails the cluster log on the management server running on `<hostname>`	2.5
check-ndbd-errlog.sh	usage: `./check-ndbd-errorlog.sh --nodeid=<id> --hostname=<hostname> --nodeid=<id> - the data node id. --hostname=<hostname> - the hostname of the data node` Prints the ndb_<id>_error.log on `<hostname>`	2.5
check-mysqld-errlog.sh check-mysql-error.sh (old name)	usage: `./check-mysqld-errlog.sh --hostname=<hostname> --hostname=<hostname> - the hostname of the mysql server` Prints the error.log from the mysql server on `<hostname>`	2.5
execute-all-mysql.sh	usage: `./execute-all-mysql.sh --user=<username> --password=<password> --database=<db> --execute=<query>` -u\|--user=<username>- the username to connect to the mysql server as. -p\|--password=<password> - the password of username -d\|--database=<database&rt; - database to connect to -e\|--execute=<query> - query to execute all parameters are optional (except --execute), default is equiv. to 'mysql -uroot test' (no password)	2.5
error-report.sh	usage: `./error-report.sh` Generates an error report (collected logs etc). Useful for filing bugs and Support issues. The error report will be stored in `error_reports` in cwd.	2.5
query-param.sh	usage: `./query-param.sh <parameter>` Prints the values of <parameter>. <parameter> can e.g be DataMemory, IndexMemory, MaxNoOfConcurrentOperations etc	2.5
sanity-check-config.sh	usage: `./sanity-check-config.sh` Verifies parameters in config.ini.	2.7

File structure and organization

Below is the organization of scripts and directories.

mysqlcluster-XX
|-- cluster
|    |-- build (used for src builds)
|    |-- repo (binary .tgz of MySQL Cluster is stored here)
|    |-- config (config.ini, my.cnf etc)
|    |-- init.d (initd scripts for cluster)
|    |-- scripts (start,stop scripts..)
|            |-- install (all the install scripts..)
|                |-- .s9s (meta data describing the cluster, which version that is currently used ('latest' points to the cluster release in 'repo' which is currently installed)
|            `-- tools (tool scripts..)

Support

Severalnines provides commercial support via a subscription service.
Contact sales@severalnines.com for more information.

Troubleshooting

Cluster does not start

Verify SELinux settings, firewalls, and /etc/hosts
"sudo: sorry, you must have a tty to run sudo" -
sudo visudo
comment out: #Default requiretty
Check that you have disk space
Verify you can ping the hosts
Verify that you have enough RAM
Verify that you have enough cores for the data nodes
Verify that you haven't installed a 64-bit version on 32-bit machines

If you still can't start MySQL Cluster, please contact support@severalnines.com

Error loading shared libraries - LD_LIBRARY_PATH not set

If you get the something like the following error message when starting the mysql client (or any other mysql tool):
/usr/local/mysql/mysql/bin/mysql: error while loading shared libraries: libmysqlclient.so.16: cannot open shared object file: No such file or directory
Then you need to set the LD_LIBRARY_PATH:
E.g:
export LD_LIBRARY_PATH=/usr/local/mysql/mysql/lib/mysql:$LD_LIBRARY_PATH
or
export LD_LIBRARY_PATH=/usr/local/mysql/lib/mysql:$LD_LIBRARY_PATH
or
export LD_LIBRARY_PATH=/usr/local/mysql/mysql/lib/:$LD_LIBRARY_PATH
or
export LD_LIBRARY_PATH=/usr/local/mysql/lib/:$LD_LIBRARY_PATH
are common LD_LIBRARY_PATHSs

[MgmSrvr] ERROR -- Error opening '.../config.ini'

If you get the following error:
[MgmSrvr] ERROR -- Error opening '/.../mysqlcluster-70/scripts/../config/config.ini', error: 2, No such file or directory
Then you have forgot to run:
sh bootstrap.sh
Make sure you have run the download, dist and bootstrap scripts.

.. Bad substitution

If you get this try to run the scripts with:
./scriptname
or
bash ./scriptname

Configurator

Selections made