GIMR Database Changes in Oracle Cluster

By -

 GIMR, or Grid Infrastructure Management Repository, was introduced in Oracle 12c. Oracle’s purpose in creating GIMR was to record cluster-related events and incidents. More precisely, GIMR is intended to store Cluster Health Monitor data inside its database.

This database contains around 20 tables, which can be listed using the following commands:

[grid@RAC1 ~]$ export ORACLE_SID=-MGMTDB
[grid@RAC1 ~]$ sqlplus "/as sysdba"
SQL> alter session set container=GIMR_DSCREP_10;
Session altered.
SQL>  select table_name from dba_tables where owner='CHM' order by table_name;
TABLE_NAME
--------------------------------------------------------------------------------
CATCHMUTILPLS_PARTITION_TMP_TBL
CHMOS_ACTIVE_CONFIG_INT_TBL
CHMOS_ADVM_INT_TBL
…
16 rows selected.

For example, using the CHMOS_SYSTEM_SAMPLE_INT_TBL table, you can obtain information about disk I/O:

select to_char(begintime, 'YYYY/MM/DD HH24:mi:ss', 'NLS_CALENDAR=persian') begintime,
       round(DISK_BYTESREADPERSEC / 1024 / 1024) as MB_READ_SEC,
       round(DISK_BYTESWRITTENPERSEC / 1024 / 1024) as MB_WRITE_SEC,
       round(DISK_NUMIOSPERSEC) as IO_PER_SEC
  from CHM.CHMOS_SYSTEM_SAMPLE_INT_TBL
order by IO_PER_SEC ;
BEGINTIME           MB_READ_SEC MB_WRITE_SEC IO_PER_SEC
------------------- ----------- ------------ ----------
1400/11/03 09:53:20          14          147        856
1400/11/01 06:58:45         815            0        858
1400/11/01 06:39:55         314            2        858
1400/11/04 09:33:00          23            3        858

Before Oracle 12c, such events were stored in files with a .bdb extension:

[grid@HostName ~]$ cd $ORACLE_HOME/crf/db/HostName
[grid@HostName HostName]$ ll *.bdb
-rw-r----- 1 root root   8192 Mar  2 10:29 crfalert.bdb
-rw-r----- 1 root root 131072 Mar  2 10:29 crfclust.bdb
-rw-r----- 1 root root   8192 Mar  2 10:29 crfconn.bdb
-rw-r----- 1 root root   8192 Mar  2 10:29 crfcpu.bdb
-rw-r----- 1 root root   8192 Mar  2 10:29 crfhosts.bdb
-rw-r----- 1 root root   8192 Mar  2 10:29 crfloclts.bdb
-rw-r----- 1 root root   8192 Mar  2 10:29 crfts.bdb

In version 12.1.0.1, it became possible to select the GIMR Database option during cluster installation (optional).
However, in 12.1.0.2, it became mandatory to configure a disk group for storing GIMR database files, and this requirement persisted through Oracle 18cR3.

Note: In all these versions, GIMR could be deleted or reinstalled after cluster setup.

Starting with 18cR5, Oracle again made GIMR installation optional (similar to 12.1.0.1). This flexibility became available when installing GI 18cR5 as a Gold Image.

Press enter or click to view image in full size

In 19c, there were no major changes — the option to skip GIMR remained.

In 21c, creating a GIMR Database is not required, and by default, Oracle prevents its installation because the default setting is “Do not use a GIMR database.”

However, if you still want to use a GIMR Database in Oracle 21c, you must create it using the Oracle RAC Database Home software.
It cannot be created from Grid Infrastructure Home in this version.

To create a GIMR Database, choose the option “Use a Local GIMR database” during the GIMR Option step.

Press enter or click to view image in full size

Next, you may specify a separate disk group for the GIMR database — while not mandatory, Oracle recommends isolating it for better management and performance.

Press enter or click to view image in full size

Unlike previous versions, after GI installation in 21c, the GIMR database is not automatically created.
As explained in the installation notes, you must use RAC Database Home to create it manually.

Press enter or click to view image in full size

If you attempt to create GIMR from Grid Infrastructure Home, you will encounter the following error:

[grid@RAC2 ~]$ /grid21c/home/bin/mgmtca createGIMRContainer -storageDiskLocation +MGMT
MGTCA-1236 : Command failed, this is not a configured Database Home.

Therefore, you must install Oracle Database Home, as shown below.

Installing Oracle Database Home using the grid user

Unusually, you must install Oracle Database Home as the grid user (to create the GIMR database).

Create the directory on all nodes:

[root@RAC2 ~]# mkdir /grid21c/GIMR
[root@RAC2 ~]# chown -R grid.oinstall /grid21c/GIMR/

Unzip the Oracle 21c installation file on one node:

[grid@RAC2 ~]$ unzip /source/LINUX.X64_213000_db_home.zip -d /grid21c/GIMR

Then run the installer:

[grid@RAC2 ~]$ cd /grid21c/GIMR/
[grid@RAC2 GIMR]$ ./runInstaller

After installation, execute the following command from Oracle Database Home:

[grid@RAC2 ~]$ /grid21c/GIMR/bin/mgmtca createGIMRContainer -storageDiskLocation +MGMT

You can check the log file here:

[root@RAC2 ~]# tail -f -n 1000 /grid21c/base/cfgtoollogs/dbca/_mgmtdb/_mgmtdb0.log

[ 2022-01-26 10:48:39.602 IRST ] Prepare for db operation
DBCA_PROGRESS : 10%
[ 2022-01-26 10:48:48.036 IRST ] Registering database with Oracle Grid Infrastructure
DBCA_PROGRESS : 14%
[ 2022-01-26 10:48:49.123 IRST ] Copying database files
DBCA_PROGRESS : 43%
[ 2022-01-26 10:49:56.794 IRST ] Creating and starting Oracle instance
DBCA_PROGRESS : 62%
[ 2022-01-26 10:55:36.452 IRST ] Completing Database Creation
DBCA_PROGRESS : 71%
[ 2022-01-26 10:58:40.486 IRST ] Executing Post Configuration Actions
DBCA_PROGRESS : 100%
[ 2022-01-26 10:58:40.489 IRST ] Database creation complete. For details check the logfiles at:
 /grid21c/base/cfgtoollogs/dbca/_mgmtdb.
Database Information:
Global Database Name:_mgmtdb
System Identifier(SID):-MGMTDB

After execution, the GIMR database will be created:

[grid@RAC2 ~]$ srvctl config mgmtdb
Database unique name: _mgmtdb
Database name:
Oracle home: /grid21c/GIMR
Oracle user: grid
Spfile: +MGMT/_MGMTDB/PARAMETERFILE/spfile.284.1094986549
Password file: +MGMT/_MGMTDB/PASSWORD/pwd_mgmtdb.271.1094986131
Start options: open
Stop options: immediate
Database role: PRIMARY
Management policy: AUTOMATIC
Type: Management
PDB name: GIMR_DSCREP_10
PDB service: GIMR_DSCREP_10
Cluster name: RAC-cluster
Database instance: -MGMTDB

Note: You cannot create other databases using the Oracle Home of the GIMR Database.

[grid@RAC2 ~]$ /grid21c/GIMR/bin/dbca
[DBT-10010] Database cannot be created in the current oracle home (/grid21c/GIMR).
Press enter or click to view image in full size

Additional Notes

The GIMR Database is a single-instance database that runs on only one node of the cluster.
If the node hosting MGMTDB goes down, the database will automatically fail over to another available node.

[grid@RAC1 ~]$ srvctl status mgmtdb
Database is enabled
Instance -MGMTDB is running on node rac2
Vahid Yousefzadeh

Written by Vahid Yousefzadeh

I have been a DBA since 2011 and I work with Oracle technology. Linkdin: linkedin.com/in/vahidusefzadeh telegram channel ID:@oracledb vahidusefzadeh@gmail.com



Comments

Post a Comment

Popular posts from this blog

Oracle 21c Enhancements for TTS Export/Import

By -
  In Oracle version 21c, new features have been introduced for transportable tablespace (TTS) export/import. This article explores these capabilities in detail. Improvement 1: Parallel Execution of TTS Export/Import In Oracle 19c, TTS export/import operations could not be executed in parallel, as demonstrated by the following: Connected to: Oracle Database 19c Enterprise Edition Release 19.0 .0 .0 .0 – Production ORA-39002: invalid operation ORA-39047: Jobs of type TRANSPORTABLE cannot use multiple execution streams. Oracle addressed this limitation in version 21c, enabling parallel execution for TTS export/import: [oracle@oLinux7 ~]$ expdp directory=drm dumpfile=TTS_DUMP%U.dmp transport_tablespaces=tbs1,tbs2 transport_full_check= y parallel= 2 Export: Release 21.0 . 0 . 0 . 0 – Production on Tue Oct 12 06 : 48 : 54 2021 Version 21.3 . 0 . 0 . 0 Processing object type TRANSPORTABLE_EXPORT/STATISTICS/TABLE_STATISTICS Processing object type TRANSPORTABLE_EXPOR...
Read more

Oracle 23ai — Speed up IMPDP Using NOVALIDATE Constraints

By -
  When executing the   IMPDP   command, some operations do not support parallel execution. Creating constraints is one such operation. Oracle must validate the current data in the table before creating a constraint, which requires a full table scan for each validation. This process can be very time-consuming. Consider the following constraints: ALTER TABLE VAHID.MYTB ADD CONSTRAINT CHECK1 CHECK (col3 > 50 ); ALTER TABLE VAHID.MYTB ADD CONSTRAINT CHECK2 CHECK (col9 IN ( 5 , 15 )); We drop the  MYTB  table and recreate it using a dump file: SQL > drop table mytb; Table dropped. W -1 Processing object type TABLE_EXPORT / TABLE / TABLE_DATA W -1 . . imported "VAHID"."MYTB" 19 GB 337830912 rows in 83 seconds using external_table W -1 Processing object type TABLE_EXPORT / TABLE / CONSTRAINT / CONSTRAINT W -1 Completed 2 CONSTRAINT objects in 208 seconds As shown, creating the two constraints ta...
Read more

Staging Tables in Oracle 23ai

By -
  As you may know, using   staging tables   is a common practice in   Data Warehouse   environments, primarily during the data loading process into the   Warehouse Database . Typically, raw data received from   source databases   is first loaded into a temporary area, referred to as   staging tables . After cleansing inconsistent and invalid data and transforming it into a standard format, the cleaned data is stored in the final operational tables. In addition to initial data loading,   staging tables   are also used for updates and are often truncated before reuse. In   Oracle 23ai , a new minor feature has been introduced for this purpose, allowing users to create tables specifically designed as   staging tables . This is achieved by adding the   FOR STAGING   clause to the   CREATE TABLE   statement. Creating Staging Tables Using the   FOR STAGING   clause, staging tables can be created as ...
Read more