{"id":73077,"date":"2017-02-09T13:17:32","date_gmt":"2017-02-09T13:17:32","guid":{"rendered":"https:\/\/www.red-gate.com\/simple-talk\/uncategorized\/policy-managed-oracle-rac-one-node-databases\/"},"modified":"2021-07-14T13:06:56","modified_gmt":"2021-07-14T13:06:56","slug":"policy-managed-oracle-rac-one-node-databases","status":"publish","type":"post","link":"https:\/\/www.red-gate.com\/simple-talk\/databases\/oracle-databases\/policy-managed-oracle-rac-one-node-databases\/","title":{"rendered":"Policy-Managed Oracle RAC One Node Databases"},"content":{"rendered":"

Oracle RAC One Node, introduced in Oracle Database 11g <\/em>Release 2 (11.2.0.2), is a single instance of an Oracle RAC-enabled database running on one node in a cluster. It lets you consolidate multiple databases onto a single cluster while providing high availability benefits of failover protection to the single instance databases.<\/p>\n

Oracle RAC One Node databases can be configured to be administrator-managed or policy-managed.<\/p>\n

Administrator-managed Oracle RAC One Node<\/strong> Database: <\/strong>The database instance executes on one of the servers from the specified list of candidate nodes. Candidate servers reside in the Generic server pool and as long as at least one server from the candidate list is available, database availability is ensured.<\/p>\n

Policy-managed Oracle RAC One Node Database: <\/strong>The database instance executes on one of the members of the server pool(s) associated with the database. To ensure availability of the database, server pools need to be configured such that a server will be available for the database to fail over to in case its current node becomes unavailable.<\/p>\n

In this article, I will discuss various options available to configure server pools for policy-managed RAC One Node databases.<\/p>\n

Environment:<\/h2>\n
    \n
  • Oracle Clusterware version: 12.1.0.2<\/li>\n
  • Oracle database version: 12.1.0.2<\/li>\n
  • Type of cluster: Flex<\/li>\n
  • Hub nodes: host01<\/strong>, host02<\/strong>, host03<\/strong><\/li>\n
  • RAC One Node policy managed database: r1pmdb<\/strong><\/li>\n<\/ul>\n

    Server Pool Configuration for RAC One Node Databases<\/h2>\n

    Server pools for RAC One Node databases can be configured as one of the following:<\/p>\n

      \n
    • Sever pool of size 1<\/li>\n
    • Server pool of size > 1<\/li>\n
    • Server pool associated with a category (Oracle Clusterware 12c onwards)<\/li>\n<\/ul>\n

      Let\u2019s discuss these possibilities one by one.<\/p>\n

      Server pool of size 1<\/h2>\n

      Server pool of size 1 can be used for RAC One Node databases by setting the minimum as well as maximum size of the server pool to 1 so that there will be only one server in the server pool. To ensure availability of the database, the IMPORTANCE of the server pool should be set higher than all other server pools in the cluster, so that, if the only server in the pool fails, a new server from the free server pool or another server pool is relocated into the server pool, as required.<\/p>\n

      Demonstration<\/strong><\/h3>\n
        \n
      • Create a server pool r1pool<\/strong> (to host RAC One Node database) of size 1:<\/li>\n<\/ul>\n
        [oracle@host02 root]$ srvctl add srvpool -serverpool r1pool -min 1 -max 1<\/strong><\/span>\r\n\r\n[oracle@host02 root]$ srvctl config srvpool -serverpool r1pool\r\nServer pool name: r1pool<\/strong><\/span>\r\nImportance: 0, Min: 1, Max: 1<\/strong><\/span>\r\nCategory: hub\r\nCandidate server names:<\/pre>\n
          \n
        • Create another server pool, pool1<\/strong>, of size 1:<\/li>\n<\/ul>\n
          [oracle@host02 root]$ srvctl add srvpool -serverpool pool1 -min 1 -max 1<\/strong><\/span>\r\n\r\n[oracle@host02 root]$ srvctl config srvpool -serverpool pool1\r\nServer pool name: pool1<\/strong><\/span>\r\nImportance: 0, Min: 1, Max: 1<\/strong><\/span>\r\nCategory: hub\r\nCandidate server names:<\/pre>\n
            \n
          • Currently, one server has been assigned to each of the Free<\/strong>, pool1<\/strong> and r1pool<\/strong> server pools.<\/li>\n<\/ul>\n
            [root@host02 ~]# crsctl stat serverpool\r\nNAME=Free\r\nACTIVE_SERVERS=host01\r\n\r\nNAME=Generic\r\nACTIVE_SERVERS=\r\n\r\nNAME=ora.pool1\r\nACTIVE_SERVERS=host03\r\n\r\nNAME=ora.r1pool\r\nACTIVE_SERVERS=host02<\/pre>\n
              \n
            • Using DBCA, create a policy managed RAC One Node database r1pmdb<\/strong> assigned to server pool r1pool:<\/strong><\/li>\n<\/ul>\n
              [oracle@host02 root]$ srvctl config database -d r1pmdb\r\n\r\nDatabase unique name: r1pmdb\r\nDatabase name: r1pmdb<\/strong><\/span>\r\nOracle home: \/u01\/app\/oracle\/product\/12.1.0\/dbhome_1\r\nOracle user: oracle\r\nSpfile: +DATA\/R1PMDB\/PARAMETERFILE\/spfile.282.929472875\r\nPassword file: +DATA\/R1PMDB\/PASSWORD\/pwdr1pmdb.286.929472177\r\nDomain:\r\nStart options: open\r\nStop options: immediate\r\nDatabase role: PRIMARY\r\nManagement policy: AUTOMATIC\r\nServer pools: r1pool<\/strong><\/span>\r\nDisk Groups: DATA\r\nMount point paths:\r\nServices: sr1pmdb\r\nType: RACOneNode<\/strong><\/span>\r\nOnline relocation timeout: 30\r\nInstance name prefix: r1pmdb\r\nCandidate servers:\r\nOSDBA group: dba\r\nOSOPER group: oper\r\nDatabase instances:\r\nDatabase is policy managed<\/strong><\/span>\r\n<\/pre>\n
                \n
              • We can see that the database r1pmdb<\/strong> is currently executing on host02, the only member of server pool r1pool:<\/strong><\/li>\n<\/ul>\n
                [root@host02 ~]# srvctl status database -d r1pmdb\r\n\r\nInstance r1pmdb_1 is running on node host02<\/strong><\/span>\r\nOnline relocation: INACTIVE\r\n<\/pre>\n
                  \n
                • As we stop CRS on host02<\/strong> where r1pmdb<\/strong> is currently running, host02<\/strong> ceases to be a part of the cluster and host01,<\/strong> which is free server, moves to r1pool <\/strong>so that database r1pmdb<\/strong> now runs on host01<\/strong> \u2013 the newly-assigned member of sever pool r1pool.<\/strong><\/li>\n<\/ul>\n
                  [root@host02 ~]# crsctl stop crs \u2013f\r\n\r\n[root@host01 ~]# crsctl stat serverpool\r\nNAME=Free\r\nACTIVE_SERVERS=\r\n\r\nNAME=Generic\r\nACTIVE_SERVERS=\r\n\r\nNAME=ora.pool1\r\nACTIVE_SERVERS=host03\r\n\r\nNAME=ora.r1pool<\/strong><\/span>\r\nACTIVE_SERVERS=host01<\/strong><\/span>\r\n\r\n[root@host01 ~]# srvctl status database -d r1pmdb\r\n\r\nInstance r1pmdb_1 is running on node host01<\/strong><\/span>\r\nOnline relocation: INACTIVE<\/pre>\n
                    \n
                  • Let\u2019s stop CRS on host01<\/strong> where database r1pmdb<\/strong> is currently running. It can be seen that host03, <\/strong>the only surviving server in the cluster,<\/strong> which is member of server pool pool1<\/strong>, does not move to server pool r1pool <\/strong>because IMPORTANCE of r1pool<\/strong> is not higher than that of pool1. <\/strong>As a result, r1pool<\/strong> becomes empty and database r1pmdb<\/strong> is no longer running on any of the nodes.<\/li>\n<\/ul>\n
                    [root@host02 ~]# crsctl stop crs \u2013f\r\n\r\n[root@host03 ~]# crsctl stat serverpool\r\nNAME=Free\r\nACTIVE_SERVERS=\r\n\r\nNAME=Generic\r\nACTIVE_SERVERS=\r\n\r\nNAME=ora.pool1<\/strong><\/span>\r\nACTIVE_SERVERS=host03<\/strong><\/span>\r\nNAME=ora.r1pool<\/strong><\/span>\r\nACTIVE_SERVERS=<\/strong><\/span>\r\n\r\n[root@host02 ~]# srvctl config srvpool -serverpool r1pool\r\nServer pool name: r1pool<\/strong><\/span>\r\nImportance: 0<\/strong><\/span>, Min: 1, Max: 1\r\nCategory: hub\r\nCandidate server names:\r\n\r\n[oracle@host02 root]$ srvctl config srvpool -serverpool pool1\r\n\r\nServer pool name: pool1<\/strong><\/span>\r\nImportance: 0<\/strong><\/span>, Min: 1, Max: 1\r\nCategory: hub\r\nCandidate server names:\r\n\r\n[root@host03 ~]# srvctl status database -d r1pmdb\r\nDatabase is not running<\/strong><\/span>.\r\nOnline relocation: INACTIVE<\/pre>\n
                      \n
                    • Let\u2019s increase IMPORTANCE of r1pool<\/strong> to 10. As a result, host03<\/strong> moves from pool1<\/strong> to r1pool<\/strong> and database r1pmdb<\/strong> starts executing on host03:<\/strong><\/li>\n<\/ul>\n
                      [oracle@host03 ~]$ srvctl modify srvpool -serverpool r1pool -importance 10\r\n\r\n[oracle@host03 ~]$ srvctl config srvpool -serverpool r1pool\r\nServer pool name: r1pool<\/strong><\/span>\r\nImportance: 10<\/strong><\/span>, Min: 1, Max: 1\r\nCategory: hub\r\nCandidate server names:\r\n\r\n[root@host03 ~]# crsctl stat serverpool\r\nNAME=Free\r\nACTIVE_SERVERS=\r\n\r\nNAME=Generic\r\nACTIVE_SERVERS=\r\n\r\nNAME=ora.pool1\r\nACTIVE_SERVERS=\r\n\r\nNAME=ora.r1pool<\/strong><\/span>\r\nACTIVE_SERVERS=host03<\/strong><\/span>\r\n\r\n[root@host03 ~]# srvctl status database -d r1pmdb\r\n\r\nInstance r1pmdb_1 is running on node host03<\/strong><\/span>\r\nOnline relocation: INACTIVE<\/pre>\n
                      Thus, if a server pool of size 1 is used to host a RAC One node database, the IMPORTANCE of the server pool should be set higher than all other server pools in the cluster to ensure availability of the database.<\/p>\n
                      Server pool of size > 1<\/h2>\n
                      We can also configure a server pool to have size > 1 for a RAC one Node database. In this case, the database will execute on any one node in the server pool. If that node leaves the cluster, the database will automatically failover to another node in the server pool. When the pool falls below its minimum size, servers from Free or other server pools in the cluster may move into the deficient pool depending upon the configuration. If it is desired that RAC one Node database runs only on certain nodes in the cluster, the associated server pool can be configured accordingly. As long as there is at least one server in the server pool, RAC one node database will be available.<\/p>\n
                      Demonstration<\/h3>\n
                        \n
                      • Let\u2019s increase the size of server pool r1pool<\/strong> to 2 and specify host02<\/strong> and host03<\/strong> as its members so that the database r1pmdb<\/strong> will execute only on host02<\/strong> and host03:<\/strong><\/li>\n<\/ul>\n
                        [oracle@host03 ~]$ srvctl modify srvpool -serverpool r1pool -min 1 -max 2<\/strong><\/span> -category \"\"\r\n\r\n[oracle@host03 ~]$ srvctl modify srvpool -serverpool r1pool -servers \"host02,host03\"<\/strong><\/span>\r\n\r\n[oracle@host03 ~]$ srvctl config srvpool -serverpool r1poolServer pool name: r1pool<\/strong><\/span>\r\nImportance: 10<\/strong><\/span>, Min: 1, Max: 2\r\nCategory:\r\nCandidate server names: host02,host03 <\/strong><\/span>\r\n<\/pre>\n
                          \n
                        • On starting CRS on host01<\/strong> and host02<\/strong>, host02<\/strong> becomes a member of r1pool<\/strong> whereas host01<\/strong> is assigned to pool1<\/strong>.<\/li>\n<\/ul>\n
                          [root@host01 ~]# crsctl start crs\r\n[root@host02 ~]# crsctl start crs\r\n[root@host01 ~]# crsctl stat serverpool\r\nNAME=Free\r\nACTIVE_SERVERS=\r\n\r\nNAME=Generic\r\nACTIVE_SERVERS=\r\n\r\nNAME=ora.pool1<\/strong><\/span>\r\nACTIVE_SERVERS=host01<\/strong><\/span>\r\nNAME=ora.r1pool<\/strong><\/span>\r\nACTIVE_SERVERS=host02 host03<\/strong><\/span><\/pre>\n
                            \n
                          • If we now stop CRS on host03<\/strong> where the database is currently executing, only one server, i.e. host02,<\/strong> is left in r1pool<\/strong>. The database automatically relocates to host02<\/strong> \u2013 the only remaining server in r1pool.<\/strong><\/li>\n<\/ul>\n
                            [root@host01 ~]# srvctl status database -d r1pmdb\r\nInstance r1pmdb_1 is running on node host03<\/strong><\/span>\r\n\r\nOnline relocation: INACTIVE\r\n[root@host03 ~]# crsctl stop crs -f\r\n\r\n[root@host01 ~]# crsctl stat serverpool\r\nNAME=Free\r\nACTIVE_SERVERS=\r\n\r\nNAME=Generic\r\nACTIVE_SERVERS=\r\n\r\nNAME=ora.pool1\r\nACTIVE_SERVERS=host01\r\n\r\nNAME=ora.r1pool<\/strong><\/span>\r\nACTIVE_SERVERS=host02<\/strong><\/span>\r\n\r\n[root@host01 ~]# srvctl status database -d r1pmdb\r\nInstance r1pmdb_1 is running on node host02<\/strong><\/span>\r\nOnline relocation: INACTIVE<\/pre>\n