7,111 views

Corosync and STONITH via vCenter

I had to make a configuration of corosync, pacemaker and stonith via vCenter working on a Ubuntu 14.04 LTS server.

Following is just a reminder for myself so I will not forget:

  • Download “VMware-vSphere-Perl-SDK”. Version 5.5.0 works for me. VMWare-vSphere-Perl-SDK
  • Extract the file: tar -zxvpf VMware-vSphere-Perl-SDK-5.5.0-1384587.x86_64.tar.gz
  • Change directory to “vmware-vsphere-cli-distrib”
  • Install: ./vmware-install.pl. You might need to install extra dependencies.
  • Run CPAN shell for the first time: perl -MCPAN -e shell
  • While you are still in the CPAN shell execute the following command “install GAAS/libwww-perl-5.837.tar.gz”. We need an older version of libwww for perl.
  • We need to edit the file “/usr/lib/vmware-vcli/VMware/share/VMware/VICommon.pm” so perl accepts unknown ssl certificates from the vcenter. (source article).
    Add the line “$ENV{PERL_LWP_SSL_VERIFY_HOSTNAME} = 0;” after the line “use Data::Dumper;”
  • Now we need to create a credentials file. For this you will need a usera ccount and a password which is allowed to reset a vmware host via vcenter.
    Execute: /usr/lib/vmware-vcli/apps/general/credstore_admin.pl add -s <ip-vcenter> -u “<username>” -p “<password>” –credstore ~/vicredentials.xml
  • Change permissions on file “vicredentials.xml”: chmod 400 ~/vicredentials.xml
  • Move to a save location mv ~/vicredentials.xml /etc/corosync/
  • Do this on both nodes.
  • Check connection to vCenter is working allright:
    Execute VI_SERVER=<ip-vcenter>  VI_CREDSTORE=/etc/corosync/vicredentials.xml HOSTLIST=”NODE1″ RESETPOWERON=0 stonith -t external/vcenter -E -S
  • Result:
    Smartmatch is experimental at /usr/lib/stonith/plugins/external/vcenter line 34.
    Smartmatch is experimental at /usr/lib/stonith/plugins/external/vcenter line 115.
    Smartmatch is experimental at /usr/lib/stonith/plugins/external/vcenter line 152.
    Smartmatch is experimental at /usr/lib/stonith/plugins/external/vcenter line 34.
    Smartmatch is experimental at /usr/lib/stonith/plugins/external/vcenter line 115.
    Smartmatch is experimental at /usr/lib/stonith/plugins/external/vcenter line 152.
    info: external/vcenter device OK.
  • When you get the OK you can continue otherwise solve your problem.
  • Add stonith resource to cluster config: crm configure
    crm(live)configure# primitive p_stonith_fence_NODE1 stonith:external/vcenter \
    params HOSTLIST=”NODE1″ VI_CREDSTORE=”/etc/corosync/vicredentials.xml” VI_SERVER=”<ip-vcenter>” RESETPOWERON=”0″ pcmk_host_check=”static-list” pcmk_host_list=”NODE1″ \
    op start interval=”0″ timeout=”120″ \
    op stop interval=”0″ timeout=”120″ \
    op monitor interval=”3600″ timeout=”300″ start-delay=”15″ \
    meta target-role=”Started”
    crm(live)configure# location l_stonith_fence_NODE1 p_stonith_fence_NODE1 -inf: NODE1
    crm(live)configure# primitive p_stonith_fence_NODE2 stonith:external/vcenter \
    params HOSTLIST=”NODE2″ VI_CREDSTORE=”/etc/corosync/vicredentials.xml” VI_SERVER=”<ip-vcenter>” RESETPOWERON=”0″ pcmk_host_check=”static-list” pcmk_host_list=”NODE2″ \
    op start interval=”0″ timeout=”120″ \
    op stop interval=”0″ timeout=”120″ \
    op monitor interval=”3600″ timeout=”300″ start-delay=”15″ \
    meta target-role=”Started”
    crm(live)configure# location l_stonith_fence_NODE2 p_stonith_fence_NODE2 -inf: NODE2
    crm(live)configure# commit
    crm(live)configure# quit
  • Watch and check if the stonith resource is nicely started: crm_mon -rf1
  • Result:
    NODE1(root@NODE1):~# crm_mon -rf1
    Last updated: Thu Jul  9 08:59:45 2015
    Last change: Thu Jul  2 03:52:35 2015 via crmd on NODE1
    Stack: corosync
    Current DC: NODE1(168303913) – partition with quorum
    Version: 1.1.10-42f2063
    2 Nodes configured
    9 Resources configuredOnline: [ NODE1 NODE2]Full list of resources:

    p_stonith_fence_NODE1        (stonith:external/vcenter):     Started NODE2
    p_stonith_fence_NODE2        (stonith:external/vcenter):     Started NODE1
    Resource Group: zabbix-cluster
    fs_data    (ocf::heartbeat:Filesystem):    Started NODE2
    virtip     (ocf::heartbeat:IPaddr2):       Started NODE2
    mysqld     (ocf::heartbeat:mysql): Started NODE2
    zabbix-server      (lsb:zabbix-server):    Started NODE2
    apache2    (lsb:apache2):  Started NODE2
    Master/Slave Set: ms_drbd_data [drbd_data]
    Masters: [ NODE2]
    Slaves: [ NODE1]

    Migration summary:
    * Node NODE1:
    * Node NODE2:

Leave a Reply

Your email address will not be published.