Wednesday, 13 June 2012

REINSTALL CRS ON A RAC ENVIRONMENT


If your crs is corrupted and you want to only re-install it without affecting your database, you can use the following steps

Environment
-----------
Nodes = 2
OS Version = RHEL 5.7
Clusterware Version = 10.2.0.5
DB Version = 10.2.0.5

Steps
------

1. On both the Nodes clean RAC init scripts

Linux:
rm -rf /etc/oracle/*
rm -f /etc/init.d/init.cssd
rm -f /etc/init.d/init.crs
rm -f /etc/init.d/init.crsd
rm -f /etc/init.d/init.evmd
rm -f /etc/rc2.d/K96init.crs
rm -f /etc/rc2.d/S96init.crs
rm -f /etc/rc3.d/K96init.crs
rm -f /etc/rc3.d/S96init.crs
rm -f /etc/rc5.d/K96init.crs
rm -f /etc/rc5.d/S96init.crs
rm -Rf /etc/oracle/scls_scr
rm -f /etc/inittab.crs
cp /etc/inittab.orig /etc/inittab

2. Kill all the crsd,evmd and cssd processes on both nodes using kill -9 command

ps -ef | grep crs
ps -ef | grep evmd
ps -ef | grep cssd

3. Remove the files in /var/tmp/.oracle/ location
rm -rf /var/tmp/.oracle/

4. Remove the file in /etc/oracle/ocr.loc
rm-rf /etc/oracle/ocr.loc

5. De-install the CRS home using Oracle universe installer
     - You can ignore this step, if you don't find the CRS installation in Universal Installer "Installed Products"
      - If you can't un-install just remove the CRS directory
          rm -rf /u01/crs/oracle/product/10.2.0/crs

6. Clean out OCR and Voting disk from one node

- Check on /etc/sysconfig/rawdevices to identify your OCR and Voting disk partition
 /dev/raw/raw1 /dev/sdb1
 /dev/raw/raw2 /dev/sdb2

 /dev/raw/raw3 /dev/sdb8
 /dev/raw/raw4 /dev/sdb9
 /dev/raw/raw5 /dev/sdb11

        - You can use the fdisk command to delete and add the partition of ocr and voting disk
- The partition number gets changed after we try to delete and add the partition in logical group, so  make a note of it and modify your /etc/sysconfig/rawdevices accordingly
        - If the partion numbers are changed, use "oracleasm scandisk" on both the nodes, to make the ASM disk visible on both the nodes
- Make sure you have correct permission for Voting disk and OCR file

To Delete a partition
--------------------
[root@coltdb01 ~]# fdisk /dev/sdb

The number of cylinders for this disk is set to 48829.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., old versions of LILO)
2) booting and partitioning software from other OSs
   (e.g., DOS FDISK, OS/2 FDISK)

Command (m for help): p

Disk /dev/sdb: 51.2 GB, 51200917504 bytes
64 heads, 32 sectors/track, 48829 cylinders
Units = cylinders of 2048 * 512 = 1048576 bytes

   Device Boot      Start         End      Blocks   Id  System
/dev/sdb1               1         239      244720   83  Linux
/dev/sdb2             240         478      244736   83  Linux
/dev/sdb4             479       48829    49511424    5  Extended
/dev/sdb5           10734       20271     9766896   83  Linux
/dev/sdb6           20272       29809     9766896   83  Linux
/dev/sdb7           29810       39347     9766896   83  Linux
/dev/sdb8             479         717      244720   83  Linux
/dev/sdb9             718         956      244720   83  Linux
/dev/sdb10           1196       10733     9766896   83  Linux
/dev/sdb11            957        1195      244720   83  Linux

Partition table entries are not in disk order

Command (m for help): d
Partition number (1-11): 1

To Add a partition
-------------------------
[root@coltdb01 ~]# fdisk /dev/sdb

The number of cylinders for this disk is set to 48829.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., old versions of LILO)
2) booting and partitioning software from other OSs
   (e.g., DOS FDISK, OS/2 FDISK)

Command (m for help): p

Disk /dev/sdb: 51.2 GB, 51200917504 bytes
64 heads, 32 sectors/track, 48829 cylinders
Units = cylinders of 2048 * 512 = 1048576 bytes

   Device Boot      Start         End      Blocks   Id  System
/dev/sdb1               1         239      244720   83  Linux
/dev/sdb2             240         478      244736   83  Linux
/dev/sdb4             479       48829    49511424    5  Extended
/dev/sdb5           10734       20271     9766896   83  Linux
/dev/sdb6           20272       29809     9766896   83  Linux
/dev/sdb7           29810       39347     9766896   83  Linux
/dev/sdb8             479         717      244720   83  Linux
/dev/sdb9             718         956      244720   83  Linux
/dev/sdb10           1196       10733     9766896   83  Linux
/dev/sdb11            957        1195      244720   83  Linux

Partition table entries are not in disk order

Command (m for help): n
Command action
   l   logical (5 or over)
   p   primary partition (1-4)

7.Check that Virtual IP's are down on both the nodes
- If vip's remained up the need to be removed using : ifconfig <device> down

8. Reinstall the CRS from node 1
     - After the install of clusterware 10.2.0.1, the services can't be brought online, since the DB and ASM instance version is 10.2.0.5. You have to apply the patchset 10.2.0.5 to clusterware to
       bring up all the services online

9. Run crs_stat -t to check whether all the services and instances are up

No comments:

Post a Comment