As most probably I would need to work on similar case, tried to test something similar in my lab.
Like my favourite character says:
Unfortunately in relation to IT it is so true, if you try, definitely something will go wrong, but that troubleshooting part is also something that we like, don’t we? 🙂
So here in my example I have an Exchange server 2010 with mailbox role that got corrupted – let’s assume it is unrepairable.
First step is to rebuild/reinstall the server, important is version of the operating system – need to be the same.
So if we were installing server on server W2k8 we need to install fresh 2008 OS. /M:RecoverServer won’t allow to restore an Exchange server than it was previously installed.
Later we need to install all prerequisites (server roles/features + office filter pack), reset old computer AD account, rename computer to have same name as the old corrupted one, and join server to the domain.
Steps that would show details how to do it all can be found here:
So, I had all in place, mounted DVD to my VM, run setup.com /M:RecoverServer aaand… got an error
I run that command 1 more time – but with the same result.
Error message was not saying a lot more, so I went to the logs, and what I have found there was:
(PID 1924, Thread 10) Task test-SetupHealth writing error when processing record of index 0. Error: Microsoft.Exchange.Management.BestPracticesAnalyzer.BPAConfigurationErrorFoundException: Exchange server is a member of a database availability group.
So apparently server cannot be a DAG member when being restored. I did some more search, and found that great blog telling what need to be done:
I logged in to the other fully working server, and did my commands (server name was ZAIC8EX1, there was only one DB in DAG: ZAICMDB01):
After that I have removed server from DAG:
And perform recovery one more time, aaand… got another error:
Living around 30 year on that world learned me two things:
If you do not know why things work in certain way in world – most probably it is all about the money, and, if somehow installing something on domain/enterprise admin account you get info that you do not have permissions – most probably it is all about running the command line with elevated permissions.
So tried one more time with those, aaand… all went good 🙂
Next step was adding server back to the DAG, but when running a command I got an error:
So did some more digging and found this forum:
And this great blog:
So seems we need to evict a member node from the cluster, so I did:
And after that tried one more time:
and this time all was successful, next phase was to add database copy to that server:
aaand… that also was successful!