3.4.4 Performing a Hot Pluggable Replacement of a Flash Disk

Starting with Oracle Exadata Database Machine X7, flash disks are hot-pluggable on both Extreme Flash (EF) and High Capacity (HC) storage servers.

Also, for Oracle Exadata Database Machine X6 and earlier, the flash devices on EF storage servers are hot-pluggable. However, for HC storage servers on Oracle Exadata Database Machine X6 and earlier systems, you must power down the storage servers before replacing the flash disks.

To replace a hot-pluggable flash disk device:

1. If necessary, prepare the disk for hot-pluggable replacement.

   Typically, you will replace a flash drive only after Oracle Exadata System Software identifies a problem and sets the device status to failed - dropped for replacement, which indicates the flash disk is ready for online replacement.

   If you need to replace a flash disk that is in another state, then you must first use the CellCLI ALTER PHYSICALDISK command with the DROP FOR REPLACEMENT clause to prepare the disk for hot-pluggable replacement.

2. Verify that the flash disk is ready for hot-pluggable replacement.

   Verify that the device status is failed - dropped for replacement.

   You can check the device status by using the CellCLI LIST PHYSICALDISK command. For example:

   CellCLI> LIST PHYSICALDISK WHERE DISKTYPE=flashdisk AND STATUS LIKE '.*dropped 
   for replacement.*' DETAIL
   
            name:               FLASH_6_1
            deviceName:         /dev/nvme0n1
            diskType:           FlashDisk
            luns:               6_0
            makeModel:          "Oracle Flash Accelerator F640 PCIe Card"
            physicalFirmware:   QDV1RD09
            physicalInsertTime: 2017-08-11T12:25:00-07:00
            physicalSerial:     PHLE6514003R6P4BGN-1
            physicalSize:       2.910957656800747T
            slotNumber:         "PCI Slot: 6; FDOM: 1"
            status:             failed - dropped for replacement
   

   Note that when one FDOM fails, the affected flash disk PCI card is considered failed and the whole card must be replaced.

3. Physically locate the flash disk device.

   Use the slotNumber information from the CellCLI LIST PHYSICALDISK command output to help identify the PCI slot containing the affected flash device.

   Also, a white Locator LED is lit to help locate the affected storage server. An amber Fault-Service Required LED is lit to identify the affected flash device.

4. Make sure that the flash disk power LED is off.

   Caution:

   Removing a flash device when the power LED is on could result in a system crash. If the status is failed – dropped for replacement but the power LED is still on, contact Oracle Support Services.
5. Remove and replace the flash disk device.