Shulou(Shulou.com)06/01 Report--

Today started the rac database of a node Times wrong!

SQL > startup mount

ORACLE instance started.

Total System Global Area 608174080 bytes

Fixed Size 1220844 bytes

Variable Size 176164628 bytes

Database Buffers 427819008 bytes

Redo Buffers 2969600 bytes

ORA-01102: cannot mount database in EXCLUSIVE mode

Google found a very well-written post, the details are as follows (I have modified it! )

Analyze the reasons:

1. In the HA system, other nodes have started instances to take up resources shared by both computers (such as bare devices on the disk array).

2. It indicates that when Oracle is abnormally closed, some resources are not released. Generally speaking, there are several possibilities

1. The shared memory segment or semaphore of Oracle is not released.

2. The background processes of Oracle (such as SMON, PMON, DBWn, etc.) are not closed.

3. The files lk and sgadef.dbf used to lock the memory have not been deleted.

The solution is as follows:

When a 1102 error occurs, you can check and troubleshoot according to the following process:

If it is a HA system, check whether other nodes have started an instance to check whether the Oracle process exists. If so, kill the process to check whether the semaphore exists. If so, clear the semaphore to check whether the shared memory segment exists. If so, clear the shared memory segment to check whether the locked memory files lk and sgadef.dbf exist. If so, delete them.

Specific methods:

First of all, although our system is a HA system, the instance of the standby node is always closed, which can be confirmed by checking the database status on the standby node.

Secondly, the database downtime is caused by the power outage of the system, and the system is restarted after being connected to the power, so we rule out the 1 or 2 points of the second possibility. The most suspicious thing is the third point.

Look in the $ORACLE_HOME/dbs directory:

$cd $ORACLE_HOME/dbs

$ls sgadef*

Sgadef* not found

$ls lk*

/ opt/oracle/product/ 10.2.0/db_1/dbs/lkSIMPLY

LkSIMPLY

Sure enough, the lk file was not deleted. Delete it.

$rm lk*

The following error is encountered when starting again, but don't worry, just go ahead and get it done.

SQL > startup mount

ORACLE instance started.

Total System Global Area 608174080 bytes

Fixed Size 1220844 bytes

Variable Size 176164628 bytes

Database Buffers 427819008 bytes

Redo Buffers 2969600 bytes

ORA-00205: error in identifying control file, check alert log for more info: (

View shared memory segment

[root@simply bdump] # ipcs-map

-Shared Memory Creator/Last-op-

Shmid owner cpid lpid

786444 root 6490 6438

819213 root 6549 6438

1409040 oracle 31502 16728

Clear shared memory segments according to ID number

Ipcrm-m 1409040

My operation here is not successful, but the implementation of the following operations on ok!

View semaphore

[root@simply bdump] # ipcs-s

Key semid owner perms nsems

0x17ff6454 360448 oracle 640 154

Clear the semaphore of oracle

[root@simply bdump] # ipcrm-s 360448

Check again to confirm

[root@simply bdump] # ipcs-s

-Semaphore Arrays-

Key semid owner perms nsems

And then query the shared memory segment is also ok!

[root@simply bdump] # ipcs-m

If the Oracle process is not closed

$kill-9

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.