Recently my customer faced the following issue, wherein after completing the X7-2 Exadata Install, Flash cache was showing different size in one of the cell node than other cells.

Everything went well with onecommand install until step 15 which had this warning:

Warning:Flash Cache size is not consistent for all storage nodes in the cluster. Flash Cache on [celadm06.test.local] does not match with the Flash Cache size on the cell celadm01.test.local in cluser /u01/app/12.2.0.1/grid

We checked flashcache size using dcli command:

[root@celadm01 linux-x64]# dcli -g cell_group -l root cellcli -e "list flashcache detail" | grep size
celadm01: size: 23.28692626953125T
celadm02: size: 23.28692626953125T
celadm03: size: 23.28692626953125T
celadm04: size: 23.28692626953125T
celadm05: size: 23.28692626953125T
celadm06: size: 23.28680419921875T ==================> Smaller flashcache than other cells
celadm07: size: 23.28692626953125T

All Flash disks were in a normal state and there was no hardware failure reported.

After investigating furter through sundiag report, I found below mismatch.

name: FD_00_celadm06
comment: 
creationTime: 2018-07-22T14:11:18+00:00
deviceName: /dev/md310
devicePartition: /dev/md310
diskType: FlashDisk
errorCount: 0
freeSpace: 0 =================================================>>>>>>>>>>>>>>>>>>>>>>>>>> freeSpace is 0
id: ***********
physicalDisk: ***********
size: 5.8218994140625T
status: normal

name: FD_01_celadm06
comment: 
creationTime: 2018-07-22T14:11:18+00:00
deviceName: /dev/md304
devicePartition: /dev/md304
diskType: FlashDisk
errorCount: 0
freeSpace: 0 =================================================>>>>>>>>>>>>>>>>>>>>>>>>>> freeSpace is 0
id: ***********
physicalDisk: ***********
size: 5.8218994140625T
status: normal

name: FD_02_celadm06
comment: 
creationTime: 2018-07-22T14:11:18+00:00
deviceName: /dev/md305
devicePartition: /dev/md305
diskType: FlashDisk
errorCount: 0
freeSpace: 0 =================================================>>>>>>>>>>>>>>>>>>>>>>>>>> freeSpace is 0
id: ***********
physicalDisk: ***********
size: 5.8218994140625T
status: normal

name: FD_03_celadm06
comment: 
creationTime: 2018-07-23T19:31:59+00:00
deviceName: /dev/md306
devicePartition: /dev/md306
diskType: FlashDisk
errorCount: 0
freeSpace: 160M =================================================>>>>>>>>>>>>>>>>>>>>>>> freeSpace 160M is not released
id: ***********
physicalDisk: ***********
size: 5.8218994140625T
status: normal

So I found the culprit 🙂 The mismatch in flash cache size was caused by freeSpace not being released on one of the flash disks (FD_03_celadm06) as we can see in the logs.

I did ask customer to recreate flashcache using following procedure.

1) Check to make sure at least one mirror copy of the extents is available.

CellCLI> list griddisk attributes name,asmmodestatus,asmdeactivationoutcome
– If reporting ‘YES’ continue to step #2

2) Manually flush the flashcache:
# cellcli -e alter flashcache all flush

In a 2nd window… Check status of flashcach flush.
The following command should return “working” for each flash disk on each cell while the cache is being flushed and “completed” when it is finished.
# cellcli -e \”LIST CELLDISK ATTRIBUTES name, flushstatus, flusherror \”” | grep FD

3) Drop Flashlog:
# cellcli -e drop flashlog all

4) Drop flashcache:
# cellcli -e drop flashcache all

5) Recreate flashlog:
# cellcli -e create flashlog all

6) Recreate flashcache:
# cellcli -e create flashcache all

7) Finally check the flashcache size to see if it’s now at correct size:
# cellcli -e list flashcache detail | grep size

Issue was resolved after dropping and recreating the flashlog and flashcache on particular cellnode. 🙂

Hope so u will find this post very useful 🙂

Cheers

Regards,
Adityanath