Steps to manually create cell/grid disks on Exadata V2 if auto-create fails during disk replacement

Asset ID:	1-71-1281395.1
Update Date:	2011-01-21
Keywords:

Solution Type Technical Instruction Sure

Solution 1281395.1 : Steps to manually create cell/grid disks on Exadata V2 if auto-create fails during disk replacement

Applies to:

Exadata Database Machine V2 - Version: Not Applicable and later [Release: N/A and later ]
Information in this document applies to any platform.

Goal

These steps will help guide you to manually create the cell and grid disks for Exadata if the auto-create happens to fail for some reason.

Solution

ALL OUTPUT IS TRUNCATED.

1. In this example we are showing the issue is with disk 07. It has been recently replaced. You confirm the status in alerthistory and notice the cell and grid disks were NOT automatically created for you.
CellCLI> list alerthistory
   10_2   2011-01-04T16:15:19-05:00   clear     "Hard disk replaced. Status        : NORMAL Manufacturer : SEAGATE Model Number : ST360057SSUN600G Size          : 600G Serial Number : E11C14 Firmware      : 0705 Slot Number   : 7 Cell Disk     : No cell disk exists Grid Disk     : No grid disks exist"

2. Next confirm the physical disk and lun are just fine, but the cell and grid disk are not created.
CellCLI> list physicaldisk
20:7        E11C14      normal
CellCLI> list lun
0_7    0_7                 normal
CellCLI> list celldisk
--no 07
CD_06_dm03cel13   normal
CD_08_dm03cel13   normal
CellCLI> list griddisk
--no 07 for all groups
DATA_CD_06_dm03cel13       active
DATA_CD_08_dm03cel13       active
...
RECO_CD_06_dm03cel13       active
RECO_CD_08_dm03cel13       active
...
SYSTEMDG_CD_06_dm03cel13   active
SYSTEMDG_CD_08_dm03cel13   active

3. You can manually create the cell disk. Run the following command to confirm the disk serial number to the lun number before starting: The serial number was listed in alerthistory.
CellCLI> list physicaldisk where physicalSerial=E11C14 detail
         name:                   20:7
         ...
         luns:                   0_7
         ...
         physicalSerial:         E11C14

4. You need the above output to match the cell disk to the lun. In this example, we see the lun is 0_7. We will use the same naming convention to create the new cell disk. Check the "list celldisk" to see YOUR naming convention:
cellcli> create celldisk CD_07_dm03cel13 lun=0_7

5. Next, query a different grid disk in order to pull the necessary information (name, size, offset). In this example, we will look at the grid disks for cell disk 06 and use them as a guide:
CellCLI> list griddisk where celldisk=CD_06_dmorlcel01 attributes name,size,offset
         DATA_CD_06_dm03cel13                 220G                32M
         RECO_CD_06_dm03cel13                 308.6875G       220.046875G
         SYSTEMDG_CD_06_dm03cel13       29.125G           528.734375G

6. Create the new grid disks in this SAME order as the above output, using the sizes shown from the previous command:
CellCLI> create griddisk DATA_CD_07_dm03cel13 celldisk=CD_07_dm03cel13,size=220G
CellCLI> create griddisk RECO_CD_07_dm03cel13 celldisk=CD_07_dm03cel13,size=308.6875G
CellCLI> create griddisk SYSTEMDG_CD_07_dm03cel13 celldisk=CD_07_dm03cel13,size=29.125G

7. Run the previous query on the new grid disks and make sure all the offsets are matching (in the third column):
CellCLI> list griddisk where celldisk=CD_07_dmorlcel01 attributes name,size,offset
         DATA_CD_07_dm03cel13                 220G                32M
         RECO_CD_07_dm03cel13                 308.6875G       220.046875G
         SYSTEMDG_CD_07_dm03cel13       29.125G           528.734375G

8. At the ASM level, the old disk was likely dropped because the disk drop timer has expired. Log into +ASM1 instance and add the new disk. Set the rebalance power higher (11) to perform faster.
sql> alter diskgroup data drop disk DATA_CD_07_dm03cel13 force;
sql> alter diskgroup data add disk 'o/*/DATA_CD_07_dm03cel13' rebalance power 11;

Do the same for RECO and SYSTEMDG from a different ASM instance (to balance the work) or wait until DATA finishes on +ASM1.

Attachments

This solution has no attachment