db_{control,listener} refreshonly always do a service restart

[lupindeterd]

Hi Biemond,

What I wanted to achieve is stop the db instance only once after the opatchupgrade, then apply the patches but for some reason the stop db is being trigger every puppet run.

``Notice: /Stage[main]/Profile::Oracledb::Non_rac/Db_control[orclcondb stop_for_patching]/ensure: ensure changed 'start' to 'stop'`

oradb::opatchupgrade {  '12.1.0.1.10':
    oracle_home                          => $oracle_home,
    patch_file                                => 'p6880880_121010_SOLARIS64.zip',
    csi_number                             => undef,
    support_id                               => undef,
    opversion                                => '12.1.0.1.10',
    user                                        => $oracle_user,
    group                                      => $oracle_group,
    download_dir                          => '/install',
    puppet_download_mnt_point => '/var/tmp/oradb_patches',
 } 
 db_control { 'orclcondb stop_for_patching':
   ensure                              => 'stop',
   instance_name                 => 'orclcondb',
   oracle_product_home_dir => $oracle_home,
   os_user                             => $oracle_user,
   refreshonly                        => true,
   subscribe                          => Oradb::Opatchupgrade[,'12.1.0.1.10'],
 }
  oradb::opatch { '21948354':
   ensure                          => 'present',
   oracle_product_home  => $oracle_home,
   patch_id                       => '21948354',
   patch_file                     => 'p22191659_121020_SOLARIS64.zip',
   clusterware                  => false,
   use_opatchauto_utility => true,
   bundle_sub_patch_id  => undef,
   user                             => $oracle_user,
   group                           => $oracle_group,
   ocmrf                           => true,
   puppet_download_mnt_point => '/var/tmp/oradb_patches',
   require                          => Db_listener[ 'orclcondb stop_for_patching'],

`

[biemond]

Hi,

it looks like if opatchupgrade also goes off, maybe opversion => '12.1.0.1.10' is not the same with it retieved.

or you have somewhere a require to Db_control[ 'orclcondb stop_for_patching']

Can you run it with --debug this will show you why Db_control[ 'orclcondb stop_for_patching'] is started

[lupindeterd]

Cursory looks seems the correct opatch is being returned. Another interesting point is if I put a another resource block e.g file {} and have this db_control subscribe to it, it's exhibiting the same behavior.

Debug: Loading facts from /var/opt/lib/pe-puppet/lib/facter/puppet_vardir.rb
Debug: oradb opatch 12.1.0.1.10
<< trimmed >>
Debug: Caching connection for https://puppserver:8140
Debug: Exec[concat_/opt/oracle/product/12.1.0.2.0/db12c/network/admin/tnsnames.ora](provider=posix): Executing check '/var/opt/lib/pe-puppet/concat/bin/concatfragments.rb -o "/var/opt/lib/pe-puppet/concat/_opt_oracle_product_12.1.0.2.0_db12c_network_admin_tnsnames.ora/fragments.concat.out" -d "/var/opt/lib/pe-puppet/concat/_opt_oracle_product_12.1.0.2.0_db12c_network_admin_tnsnames.ora" -l -t'
Debug: Executing '/var/opt/lib/pe-puppet/concat/bin/concatfragments.rb -o "/var/opt/lib/pe-puppet/concat/_opt_oracle_product_12.1.0.2.0_db12c_network_admin_tnsnames.ora/fragments.concat.out" -d "/var/opt/lib/pe-puppet/concat/_opt_oracle_product_12.1.0.2.0_db12c_network_admin_tnsnames.ora" -l -t'
Debug: instance_status /bin/ps awwx | /bin/grep -v grep | /bin/grep 'ora_smon_orclcondb'
Notice: /Stage[main]/Profile::Oracledb::Non_rac/Db_control[orclcondb stop_for_patching]/ensure: ensure changed 'start' to 'stop'
Debug: /Stage[main]/Profile::Oracledb::Non_rac/Db_control[orclcondb stop_for_patching]: The container Class[Profile::Oracledb::Non_rac] will propagate my refresh event
Info: search for patchid 21948354
Info: opatch_status for patch 21948354 command: /opt/oracle/product/12.1.0.2.0/db12c/OPatch/opatch lsinventory -patch_id -oh /opt/oracle/product/12.1.0.2.0/db12c -invPtrLoc /var/opt/oracle/oraInst.loc
mesg: cannot change mode

This code snippet however works as expected.

  oradb::opatchupgrade { '12.1.0.1.10':
    oracle_home               => $oracle_home,
    patch_file                => 'p6880880_121010_SOLARIS64.zip',
    csi_number                => undef,
    support_id                => undef,
    opversion                 => '12.1.0.1.10',
    user                      => $oracle_user,
    group                     => $oracle_group,
    download_dir              => '/install',
    puppet_download_mnt_point => '/var/tmp/oradb_patches',
   notify                    => Exec[ 'stop_db_for_patching' ],
 }
 # We'll resort to this for now until we got the db_control sorted.
 exec { 'stop_db_for_patching':
   command      => "${oracle_home}/bin/sqlplus /nolog <<-EOF
                   connect / as sysdba
                   shutdown immediate
                   EOF",
   environment  => [ "ORACLE_HOME=${oracle_home}", "ORACLE_SID=orclcondb",
                    "LD_LIBRARY_PATH=${oracle_home}/lib"],
   logoutput    => true,
   user         => $oracle_user,
   onlyif       => "/bin/ps -ef | /bin/grep -v grep | /bin/grep 'ora_smon_orclcondb'",
   refreshonly  => true,
 }

[johmicd]

you might want to use an unless with opatch_version and/or oradb::is_oracle_patch_installed'

[pmpetit]

Hello, be carefull the last opatch version for a 12.1 db is a 12.2.0.12 you can see it in Version file inside the zip ... so if you install with opatchupgrade.pp you must set opatch version to 12.2.0.12 even if your db is 12.1