CRS-1006: No more Members to consider

Http://inthirties.com: 90/thread-1359-2-1.html
This Solaris migration from a single instance to RAC is not very smooth. Hehao, the holiday is over, and the younger daughter is at home, and there is more fun in the house. I want you
Give her scissors and ask you to paint and make a hand for him. This is intermittent. The database and instance have been registered on the node. You have also tried
Once again, the entire RAC has been well done and can be used. At, I took my little daughter to the children's playground in wanjia. Come back in the evening. The SSH connection fails, and the machine becomes faulty again. View servers
It's okay that the VM server is suspend. This environment is installed in the VM of the server. It may be a little girl's rush. Hide
The default option is suspend, so this is suspend. Then, after resume, the VM lost its response and did not know whether it was a VM bug,
This has happened many times on my machine. This time it was not spared, so we can start it after shutdown.

Restart the machine, crs_stat-T, and find the problem.

The status of instance and database is incorrect. Displayed

Name type target State host

--------------------

Ora. soldb. DB application online offline

Ora .... B1.inst application online offline

Obviously, CRS has started these two applications, so traget is online, but the startup fails.
So the state is still offline.

Automatic CRS startup seems to be unsuccessful.

Start it manually and try again. It still does not appear when you witness the miracle. Still failed.

However, the startup failure information is displayed.

CRS-1006: No more Members to consider

CRS-0215: cocould not start resource 'ora. soldb. db '.

It seems that there is a problem with OCR registration. In the past, after changing the IP address of the VIP address, the VIP address can be re-registered. Is that true,
According to the previous experience in VIP processing, but it does not use modify, the simple process is to remove it first, and then add it in.

Srvctl remove database-D soldb

Srvctl remove instance-D soldb-I soldb1

Srvctl add database-D soldb-o/u01/APP/Oracle/product/10.2.0/db_1

Srvctl add instance-D soldb-I soldb1-N Solaris

Start the two applications immediately and report an error, as before.

This hit my attention. I thought that this would allow my child to watch TV immediately. It seems that I should analyze it more carefully.

Startresource error for ora. soldb. soldb1.inst error code = 1

Start of 'ora. soldb. soldb1.inst 'on member 'solaris' failed.

CRS-1006: No more Members to consider

Startresource error for ora. soldb. DB error code = 1

There is no special information. Since the startup is unsuccessful, you have to check the instance logs. Other logs do not need to be viewed. It has nothing to do with this.

The alert log shows no visible information or even startup information. That is to say, the database is not connected to nomount. If nomount exists
Make sure that alert. log is updated. So. Locate the problem in registration.

Srvctl config database-D soldb-

Check the registration information of the database. The database name and spfile are empty, and add-N to srvctl.
-P parameters. Let's take a look at config. This is a good time. Both dB name and spfile have been seen.

Let's start it again. There is no miracle.

However, the problem should be fixed at the instance startup and registration.

Search for SQL plus immediately, and start up, prompting you That the familiar spfile cannot be found.

The spfile is not there, but in crs_stat, ASM exists clearly.

Is the disk in ASM not mounted successfully?

Oracle_sid = + ASM1 enter the ASM instance,

SQL> select path, mount_status from V $ asm_disk;

They are all in the close state, that is, the disk does not have a mount. In this case, of course, the spfie is not found. Spfile exists in ASM
Disk, so it certainly fails to start the instance, because nomount is not successful, so the instance's Alert Log File certainly does not have any records. Alert is nomount
Logs can be logged only when they are enabled.

Check the asm_diskgroups settings of ASM. asm_diskgroups is empty, so the ASM instance will not mount all the disks when it is started. Check
Query
In the asm_diskgroup view, the status of DG1 is still nomount. The problem is fixed. Solution: Specify the asm_diskgroups parameter. However, note that,
Since ASM is started with pfile, alter system set
Asm_diskgroups only modifies memory and will become invalid when ASM is started next time. This is why the instance cannot be started after I restarted the machine.
It should be to modify the pfile file of ASM. Here, the pfile of ASM is under $ oracle_base/admin/+ ASM/pfile/init. ora. You
Check that there is an init + asm1.ora file under $ ORACLE_HOME/dbs/. However, this file is a link and the source address is
In $ oracle_base/admin/+ ASM/pfile/init. ora

[Oracle @ Solaris ~] Ls-Al $ ORACLE_HOME/dbs/init + asm1.ora

Lrwxrwxrwx 1 Oracle oinstall 41 Apr 3 21:48
/U01/APP/Oracle/product/10.2.0/db_1/dbs/init + asm1.ora->
/U01/APP/Oracle/admin/+ ASM/pfile/init. ora

Restart the ASM instance, and then start ora. soldb. DB, tail
+ 100f open the alert file of the instance and see the startup log of the alert file. The problem should be solved.

The database instance is successfully started.

Looking back at this fault, it is true that the shortest method is used when migrating This RAC. The minimum start time of ASM is used. In ASM, the data files and other files of the previous standalone instance are used.
Manually register to RAC. After registration, use alter system set
Asm_diskgroups manually mount the DG, and then neglected that the ASM was started by pfile, resulting in this problem.

However, the error number provided here in Oracle is indeed a little different, and I am not given a clear direction, causing myself to doubt that it is srvctl.
Add. Another Oracle should provide is that when I haven't started the spfile instance to nomount, where can I find the failure log here.

First of all, nomount fails. It may be because the instance alert should have no information, and it is obviously irrelevant to cssd, so ocssd. log does not have any information.
Crsd is the registration and notification information for each node resource in CRS. It does not seem to be related to this.
The alert information of CRS is the running log of CRS, and there should be no instance startup failure information here, So/var/adm/message will probably not start the instance
The feedback of the dynamic command. This is the OS log. So the information such as this, Oracle, where do you actually record. Should you give us an explanation. Is it wrong?
Exclusion can improve efficiency.

• Provides Oracle management/troubleshooting/optimization/installation/RAC/Backup Recovery technical services, and professional Oracle training and consulting services.
  
• Mail: inthirties@gmail.com
• MSN: inthirties@hotmail.com
• QQ: inthirties@qq.com
• Tel: 13828706466
• Technical blog http://blog.csdn.net/inthirties
• Personal site http://blog.inthirties.com