Overview

You purchased a new Storage cell for expansion or removed Storage from an existing Cluster and want to add it to the new a Cluster due to space demand.

In this article I will demonstrate how to add a storage cell to an existing Exadata Database Machine.

Here we are adding new storage cell with 600 GB disks to existing Exadata database machine using 600 GB disks.

• Exadata Database Machine X2-2 Full Rack
• Exadata Storage Software version 12.1.2.2.0
• Oracle Grid/Database Version 11.2.0.4

Pre Checks:

• Create a new file named “new_cell_group” containing only the hostnames for the new cells.

• It is assumed that the file “cell_group” only contains the original cells, not the new ones.  If you have already modified “cell_group” to contain the new cells, just comment it out or remove it from the file.

• Here we use the root user to run the cellcli commands; ensure that user equivalency setup for root user between Storage cells and Compute nodes.

• First verify if any cell disks already exist:

• If no cell disks are created. There will be no output.

• If this command returns a list of cell disks, then you should verify if any grid disks exist:

• If no grid disks are created. There will be no output.

• If any grid disks are present, then you must ensure they are not in use by looking in V$ASM_DISK and verifying the mount status is “closed” and the header status is “former” or “candidate”:

• If it is OK to drop the grid disks, you can use the following command to drop set of grid disks (by prefix like, RECO):

• Verify Grid disks are dropped

• Drop cell disks

• If no cell disks exist, create cell disks on all hard disks (output will be similar to the following): 

• Verify that celldisks were created (output should be similar to the following)

• Verify the current sizes and offsets for existing grid disks (on the existing cells):

• Create similar grid disks on the new cells with the same size; be sure to create them in order of offset position, from smallest to largest offset. 

• Validate that grid disks were created properly on the new cells (size and offset matches the original cells)

• List the size of the DBFS_DG grid disks on the original cells (they should all be the same size) – we’ll call this size, “dbfs_dg_size”:

• The DBFS_DG grid disks will be automatically placed just after the RECO grid disks on the innermost tracks.

• All Cells at once:

• One cell at a time:

• Update the cellip.ora file

• From one DB node add the new DATA disks to the DATA disk group (replace the pattern below ” ‘o/*/DATA_DM01*new_cel01’, ‘o/*/DATA_DM01*new_cel02’, ‘o/*/DATA_DM01*new_cel03′” with list of actual new cells):

•  From another DB node add the new RECO disks to the RECO disk group (replace the pattern below, ” ‘o/*/RECO_DM01*new_cel01′,’o/*/RECO_DM01*new_cel02′,’o/*/RECO_DM01*new_cel03’ ” with list of actual new cells):

• Monitor the progress of the rebalance activities using this query (you can proceed with the rest of the steps while this is continuing):