On Fri, Aug 2, 2013 at 10:18 AM, Adrian Klaver <adrian.klaver@xxxxxxxxx> wrote:
If by "modifying the registry," Mr. Brearley means following the procedure described here on the PostgreSQl Wiki, then that actually will change the data directory. That procedure actually modifies the Windows service command line (stored in the registry), and it changes the -D argument. (Yes, that appears to have been written against 8.3RC2, but I checked, and it is still applicable to 9.2 if names are changed appropriately.)
Instead of changing the service, it might be more effective to just stop the service and manually start PostgreSQL from the command line until the data can be retrieved. That would allow for somewhat more consistent usage with other operating systems, making it simpler to help. If he can get something to connect to that data directory, that would allow him to at least dump the data so it could be imported into a new, empty cluster.
My big concern as I've been reading this thread is whether users are cluster specific or installation specific. If they're cluster specific, he'll need to know credentials for his original cluster anyway to get the data, unless he can do some kind of password reset.
You can't do that and have it work. \base is just part of the puzzle, you need the complete \data directory for Postgres to work.The problem(as stated before) is you have two distinct installations of the Postgres data directory, one on D:\ and one on C:\. In Postgres parlance the \data directory in each of those locations represents a database cluster. The procedure you described above for the successful install set up Postgres to run off the C:\ cluster. All the stuff you are doing on D:\ will be of no use until you point Postgres at the cluster on D:\(I don't think changing the registry entry counts). Unfortunately my lack of knowledge regarding Postgres on Windows means I will be of no help there. Someone else will need to fill in the procedure.
I have to ask again, how important is the old data and how much is there?
Would it be easier to just reenter it in the new cluster?
If by "modifying the registry," Mr. Brearley means following the procedure described here on the PostgreSQl Wiki, then that actually will change the data directory. That procedure actually modifies the Windows service command line (stored in the registry), and it changes the -D argument. (Yes, that appears to have been written against 8.3RC2, but I checked, and it is still applicable to 9.2 if names are changed appropriately.)
Instead of changing the service, it might be more effective to just stop the service and manually start PostgreSQL from the command line until the data can be retrieved. That would allow for somewhat more consistent usage with other operating systems, making it simpler to help. If he can get something to connect to that data directory, that would allow him to at least dump the data so it could be imported into a new, empty cluster.
My big concern as I've been reading this thread is whether users are cluster specific or installation specific. If they're cluster specific, he'll need to know credentials for his original cluster anyway to get the data, unless he can do some kind of password reset.
