Saturday, January 12, 2013

Shutting Down a Exadata DB Node [same on normal RAC]


While we need to shutdown any of exadata(same for any RAC DB) DB Nodes I normally follow below checklist. 

1. from DB & ASM node run below query and cahek all the nodes are in open staus
select * from gv$instance;
2. login to the target DB node with oracle user and set .oracrs.sh env

3. now shutdown the instance on the target node using below command
/u01/app/11.2.0/grid/bin/srvctl stop instance -d ORADB -i

4. FROM DB node run belwo quert to be sure that target instance is now down
select * from gv$instance;

5. now login as root and run below command to shudown CRS
/u01/app/11.2.0/grid/bin/crsctl stop crs

-- from DB node run below command and make sure that crs is down [CRS-0184]
/u01/app/11.2.0/grid/bin/crs_stat -t
CRS-0184: Cannot communicate with the CRS daemon.

6. FROM ASM run below query to make sure that ASM on tgaret hots is down
select * from gv$instance;

7. shutdown the DB machine

8. startup the DB machine

9. check with below command the all the cluster services are online except .gsd and .oc4j
/u01/app/11.2.0/grid/bin/crs_stat -t|grep OFFLINE
--from DB node with oracle user and .oracrs.sh env run below command to make sure that all the ASM disks are online
asmcmd
ASMCMD> ls

. oracrs.sh
sqlplus / as sysasm

10. if any disk is missing the from ASM instance of the target machine run below command to make the perticular disk online
alter diskgroup mount;

11. chek that DB instance is online or not if not automatically online use below command to make it online
srvctl start -d axdw -i

12. from ASM & DB instance user below query to check all the ASM and DB instances are online
select * from gv$instance;

13. from ASM & DB instance user below query to check all the ASM DISKS are online from all the instance
select header_status,mode_status,state,failgroup,count(*) from gv$asm_disk group by header_status,mode_status,state,failgroup;

14. check alertlog of both db & asm for the target node

No comments:

Post a Comment