Wednesday, April 13, 2016

Klaster mariaDB Centos 7 OS

mysql -u root -p -e "show status like 'wsrep%'"
mysql -u root -p -e "show status like 'wsrep_local_state_comment%'"


5 comments:

  1. Scenario kako se vraća klaster
    https://www.percona.com/blog/2014/09/01/galera-replication-how-to-recover-a-pxc-cluster/

    ReplyDelete
  2. removed galera.cache and the log files in /var/lib/mysql/

    It seems that i changed the log file size in my.cnf, that's why it won't start.
    That's weird, because i've seen this error before, and it's not what i had in the logs...
    Anyway it's working.

    CODE: SELECT ALL
    rm -f /var/lib/mysql/galera.cache ib_*


    I have also rebooted.

    ReplyDelete
  3. IF EVERY NODE SHUTDOWNS

    check grastate of each node, the highest seq. is the latest node you should initialize from

    $ sudo tail -100 /var/lib/mysql/grastate.dat
    initialize the highest node

    $ sudo service mysql start –wsrep-new-cluster

    if you see -1 in each node’s seqno you can recover the node by running

    $ sudo /usr/sbin/mysqld -u root –wsrep-recover

    hard to know which node has the latest when all nodes have -1, then you need to make a guess and must pick up the node and initialize the cluster

    $ sudo service mysql start –wsrep-new-cluster



    IF TWO NODES SHUTDOWN, ONE IS OK

    try first to make join the 2 nodes to the existing cluster one by one

    $ sudo service mysql start

    if doesn’t start, try to recover each node to get its position internally

    $ sudo /usr/sbin/mysqld -u root –wsrep-recover
    $ sudo service mysql start

    then run other nodes the same way

    $ sudo service mysql start
    Starting MySQL…SST in progress, setting sleep higher………………………………………….. SUCCESS!



    TURN OF a SINGLE NODE

    let’s turn off a single node, start from the latest i.e. db3

    $ sudo service mysql stop

    to make sure all mysql processes are shutdown

    $ ps -ef | grep mysql

    or

    $ sudo pidof mysqld

    if any mysql still running, then kill it

    $ sudo kill -9 pid

    on db2 or db1 check out cluster size

    mysql>show status like ‘wsrep%’;



    STOP THE ENTIRE GELARA CLUSTER (NOT RECOMMENDED)

    mysql>show status like ‘wsrep_local_state%’;

    if Synced then it is OK to start the shutdown node procedure. If the node is non-synchronized, you may shut it down anyway, but avoid to start a new cluster from this node in the future

    mysql> set global wsrep_on=‘OFF’

    ReplyDelete
  4. Lepo opisano kako se instalira mariadb na centos 7
    https://snapdev.net/2015/07/21/mariadb-cluster-on-centos-7/

    ReplyDelete