Marin Komadina
Cluster Configuration
On a Sun Ultra-Enteprise-10000 machine,
with 2 CPU, 100MHz, and 4096 Megabytes RAM, we have installed a Solaris 2.8
operating system and cluster software Sun Cluster 2.2.
One active Oracle database (artist000)
running in failover configuration with HA Oracle Agent exists on that cluster.
Sun Cluster check
Cluster administration is usually under the
control of the UNIX administrator. As a DBA, we have only to check if the
cluster is running or not. If not, then it has to be started with scadmin command as root.
$ haoracle list
Cluster is not up; Run scadmin {startcluster|startnode}first.
scadmin startcluster CNODEB CLUST00C2
When the cluster is running, we can retrieve basic cluster
information:
root@cnodea# haget -f mastered
CNODEA
CNODEB
Cluster
configuration has two member nodes, CNODEA and CNODEB.
To find out more detailed information we can use command hastat:
cnodeb@root#hastat
Getting Information from all the nodes ......
HIGH AVAILABILITY CONFIGURATION AND STATUS
-------------------------------------------
LIST OF NODES CONFIGURED IN <CLUST00C2> CLUSTER
CNODEA CNODEB
CURRENT MEMBERS OF THE CLUSTER
CNODEA is a cluster member
CNODEB is a cluster member
CONFIGURATION STATE OF THE CLUSTER
Configuration State on CNODEB: Stable
Configuration State on CNODEA: Stable
UPTIME OF NODES IN THE CLUSTER
uptime of CNODEB:
12:07pm up 10 day(s), 5:14, 4 users, load average: 1,10, 0,60, 0,42
uptime of CNODEA:
12:07pm up 10 day(s), 5:14, 1 user, load average: 0,10, 0,15, 0,17
LOGICAL HOSTS MASTERED BY THE CLUSTER MEMBERS
Logical Hosts Mastered on CNODEB: LOGNODEH
Logical Hosts for which CNODEB is Backup Node: None
Logical Hosts Mastered on CNODEA: None
Logical Hosts for which CNODEA is Backup Node: LOGNODEH
LOGICAL HOSTS IN MAINTENANCE STATE
None
STATUS OF PRIVATE NETS IN THE CLUSTER
Status of Interconnects on CNODEB:
interconnect0: selected
interconnect1: up
Status of private nets on CNODEB;
To CNODEB - UP
To CNODEA - UP
Status of Interconnects on CNODEA:
interconnect0: selected
interconnect1: up
Status of private nets on CNODEA:
To CNODEB - UP
To CNODEA - UP
STATUS OF PUBLIC NETS IN THE CLUSTER
Status of Public Network On CNODEB :
bkggrp r_adp status fo_time live_adp
nafo0 qfe0 OK NEVER qfe0
Status of Public Network On CNODEA :
bkggrp r_adp status fo_time live_adp
nafo0 qfe0 OK NEVER qfe0
STATUS OF DATA SERVICES RUNNING IN THE CLUSTER
Status Of Registered Data Services
oracle: On
Status Of Data Services Running On CNODEB
Data Service "oracle":
Database Status on CNODEB :
artisit000 - running;
Status Of Data Services Running On CNODEA
Data Service "oracle":
Not being managed on this system
Physical node CNODEB is now
mastering logical host LOGNODEH. The Oracle database instance artist000
is running in cluster CLUST00C2, on physical node CNODEB.
Volumes associated with the existing database:
/LOGNODEH
/oracle/app/oracle/admin/artist000
/oracle/app/oracle/admin/archspace/arch
/oracle/dataspace/artist000
/LOGNODEH - A special volume which holds cluster
statistical information, and is usually few MB in size
/oracle/app/oracle/admin/artist000 - Volume group holds
configuration and initialization database files
/oracle/app/oracle/admin/archspace/arch - Volume group holds
database archived log files
/oracle/dataspace/artist000 - Volume group holds database
files
HA Oracle agent check
To
find out the actual configuration for Oracle services in a cluster, we can
check the cluster configuration (hastat):
# hastat
--
Status Of Data Services Running On CNODEB
Data Service "oracle":
Database Status on CNODEB:
artist000 - running;
--
or using HA Oracle agent (haoracle)
command
#haoracle list
on:artist000:LOGNODEH:60:10:120:300:ha/test:/oracle/app/oracle/
admin/artist000/pfile/initartist000.ora:LISTENER_artist000
All actions made by HA Oracle
Agent software are logged on location /var/opt/SUNWscor in file hadbms.log,
where we can search for additional activity information.
Oracle
Configuration Check
In a clustered system, Oracle
installation can be on a shared volume or on a local disk.
On our test system, we have two nodes and
Oracle binaries have been installed on each physical host local disk. Oracle HA
Agent uses a special directory on location /var/opt/oracle, with Oracle
initialisation and configuration files.
-rw-rw-r-- 1 oracle dba 439 Feb 6 10:52 oratab
-rwxr-xr-x 1 oracle dba 4320 Feb 6 10:52 listener.ora*
-rw-r--r-- 1 oracle dba 1216 Feb 6 10:53 tnsnames.ora
Oratab is a special
configuration file with entries for all
of the database instances on all nodes running HA Oracle Agent software. The
Oracle DBA must manually keep this file current on all nodes to successfully
failover.
#more oratab
# Y/N Y0 enable dbstart and dbshut Scripts
artist000:/oracle/app/oracle/product/8.1.5:N
All entries in the oratab file should have the :N option
specified to ensure that the instance will not start automatically on machine
reboot.
Listener.ora
Listener configuration file entries define the communication protocol and settings for database connections.
LISTENER_artist000 =
( ADDRESS_LIST =
( ADDRESS = ( PROTOCOL = TCP ) ( Host = LOGNODEH)(Port= 1528))
( ADDRESS = ( PROTOCOL = IPC ) ( KEY = artist000 ) ))
SID_LIST_LISTENER_artist000 =
(SID_LIST = (SID_DESC =
(ORACLE_HOME = /oracle/app/oracle/product/8.1.5)(SID_NAME = artist000)))
Tnsnames.ora
This is the Oracle network connection configuration file, with
defined "service names" for every instance and logical host defined in listener.ora file.
artist000 =
( DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = LOGNODEH)(PORT = 1528))
(CONNECT_DATA = (SID = artist000)))