How could we start Oracle database in Oracle Exadata which's celldisk has proactive failure?

Question

I am trying to open exadata machine but it has error. That EXADATA Oracle Sun Fire X4720 M3 SAS, Oracle 11gr2 RAC database installed. I was examining celldisks. When i gave command list griddisk I see that result.

cellcli> list griddisk
  DATA_DM01_00_mycelldisk01 proactive failure

We dropped celldisk, we can't recreated griddisk and celldisk, it has error. we can't replace it because we can't get any disk from anywhere. I need to open that exadata and i plan to open database without that celldisk. should i drop physicaldisk?

cellcli> list physicaldisk
  20:0 KERMXM warning - proactive failure, poor performance

when I give command srvctl start database -d mydb it gives error.

ORA-15040: diskgroup is incomplete
ORA-15066: offlining disk "0" in group DATA_DM01 may result in a data loss
ORA-15042: ASM disk "0" is missing from group number "1"

Could i delete disk "0" which is proactive failure and continue my system? Thanks

Answer 1

Exadata uses NORMAL or HIGH redundancy diskgroups. It is possible to use EXTERNAL redundancy diskgroups as well, but that is not supported.

Even if you drop a griddisk, and one of the ASM failgroups becomes incomplete, you can still mount the ASM diskgroup and use the mirror side(s). But this will not happen automatically, because the usual MOUNT command throws an error in such a case, and srvctl will not start the database because of that.

You should be able to mount the diskgroup manually with the FORCE option. Log in to the ASM instance as sysasm:

sqlplus / as sysasm

And mount the diskgroup (on all nodes):

alter diskgroup data_dm01 mount force;

Once the diskgroup is mounted, you should be able to start the database.

For the faulty physicaldisk, you should open an SR with Oracle (given that you have valid support contract) for a replacement drive (this usually automatically happens with ASR (Auto Service Request)). The last time I did that, the replacement drive was available in 1-2 days, and their engineer delivered and replaced it in the customers data center.

• Melih: it gives same error. there is no support license so we can not call