tychos wrote:This is a great reply. Thanks for taking the time - it's very much appreciated.
You can reinstall your V2 without losing support. (I got this confirmed after raising an SR at MOS.)
I did this exercise 3 months ago coming from a similar type of situation.
I had to choose to upgrade from 18.104.22.168.2 to 22.214.171.124.1 and from 126.96.36.199 to 188.8.131.52 and change the RECO to DATA ratio and change some IPs
do a reinstall.
Points of concern:
- You have to have a valid CSI for Exadata software as the password of the Exadata firmware images (db and cell) are password protected. (I think you have this)
- Next to that you have to reverse engineer the factory settings to be able to use onecommand successfully and run an automated install. (This is not documented)
- You need to like to puzzle. (I guess you do)
we'll have to upgrade the O/S (5.5 to 5.7)The OS on Exadata is not a standard iso image used on any other server or VM.
Is the reinstall-by-yourself method 'approved' by Oracle?According to support it is.
would they respond to a SR or would they tell me to contact ACS instead of bothering them?I didn't need them for the reinstall so I can't tell how they would respond in this situation.
tychos wrote:Thanks, again, for your input. I have restored a storage cell using the USB drive, but that was to the same version as was running currently.
Hi Mark,we'll have to upgrade the O/S (5.5 to 5.7)The OS on Exadata is not a standard iso image used on any other server or VM.
(It will be more clear if you read the bare metal restore procedure mentioned below)Is the reinstall-by-yourself method 'approved' by Oracle?According to support it is.
As an example: If you loose a db-node you can reimage it by the "reinstall-by-yourself" method.
Check out DocID 1084360.1 Bare Metal Restore Procedure for Compute Nodes on an Exadata Environment.
This is BTW a good starting point to understand a full rebuild.would they respond to a SR or would they tell me to contact ACS instead of bothering them?I didn't need them for the reinstall so I can't tell how they would respond in this situation.
In general they are willing to help out.
frits hoogland wrote:I'm fairly sure that we would be able to upgrade - in fact, when we weren't entertaining a hardware upgrade earlier in the year, I had planned out a upgrade path from our current versions to what was the current stack before the FlashCache became write-able.
I don't understand the answers.
A V2 Exadata system (and up, X2, X3) is full supported up to the newest Exadata software releases, so you just can upgrade. Of course you need to check with MOS 888828.1 what path to take (not all software might be upgradable to the latest release in one go). Need need to puzzle, just upgrade.
If you want to change the space ratio between DATA and RECO, the easy path is to delete all the databases, remove the data and reco diskgroups, remove the grid disks on the cells, and create them again, and create the diskgroups on top of it. This also can be done online by dropping the griddisks per cell/storage server in ASM, recreating them with different sizes, and get them in ASM again.I believe that Tycho said he had to choose between upgrading the stack AND change the space ratio between his diskgroups OR just rebuild the system from scratch: and he chose to rebuild.