Buy Percona ServicesBuy Now!

New 5.7 cluster, 3 CentOS 7 nodes don't join

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • New 5.7 cluster, 3 CentOS 7 nodes don't join

    Setup a 3 node cluster starting here: https://www.percona.com/doc/percona-...l/yum.html#yum

    all three VM's are fresh centos 7 installs.

    Node 1 I goofed up and had to uninstall, remove the data dir, and reinstall percona cluster completely.

    before I bootstrapped, I had to set strict mode to disable to import my .sql file of the data.

    Turned strict mode back on. started bootstrap.
    Turned on Node 2.
    It never joined.

    Node 3, same. They all show cluster size of 1.

    I then read that if i have to stop the bootstrap service, I should just restart using systemctl start mysql.service. So I did that (because i had restarted the bootstrap service) but node 2 and 3 still don't join.

    I'm not sure what i've missed but this thread I've linked, is the exact problem i'm having.

    Node 2 Log: https://pastebin.com/3G8WYK8Y

    What I can't seem to find is, how do i tell node 2 to forget anything it had and join the cluster? It can't be a reinstall.

    firewalld has been disabled on all 3 servers. SELinux is also disabled. This a test to learn about percona. All three can ping each other and are on the same subnet.
    I've been trying to set up a 3-node XtraDB Cluster. I followed these steps: https://pastebin.com/LHR57aYT The problem is that the second node does not even

  • #2
    The error log you shared shows a single node being bootstrapped, it did not connect to any peer. You should bootstrap only the first node, the others should join him via normal service start.
    Please show the configs you used on those nodes.

    Comment


    • #3
      Originally posted by przemek View Post
      The error log you shared shows a single node being bootstrapped, it did not connect to any peer. You should bootstrap only the first node, the others should join him via normal service start.
      Please show the configs you used on those nodes.
      But that isn't what happened. I bootstrapped node 1, then started node 2 normally. It never joined and had "Synced" as the status.

      Node 1:https://pastebin.com/LYiBCyZw
      Node 2:https://pastebin.com/0wQvSEZ1
      Node 3: https://pastebin.com/ptyyP0uN

      Node 1, Node 2, Node 3 all report identically:
      wsrep_local_state_comment | Synced
      wsrep_cluster_conf_id | 1
      wsrep_cluster_size | 1

      Comment


      • #4
        The issue was that the documentation is either out of date or just wrong.
        Mysql was loading each of my nodes in "bootstrap" mode because of /etc/percona-xtradb-cluster.conf.d/wsrep.cnf, this file was being loaded last and overwriting my entry for wsrep_cluster_address.

        After i discovered this, I updated wsrep.cnf to match what the instructions said needed to be set. The nodes now see each other EXCEPT, as soon as node 2 tries to start it changes node 1 from Synced to Initialized. Node 2 mysql fails to start with an error. I'll start a new topic

        Comment


        • #5
          Thank you for pointing this out. I have reported this problem under https://jira.percona.com/browse/PXC-2258

          Comment

          Working...
          X