On 12/19/2013 01:50 PM, Joseph Kregloh wrote:
On Thu, Dec 19, 2013 at 4:14 PM, John R Pierce <pierce@xxxxxxxxxxxx <mailto:pierce@xxxxxxxxxxxx>> wrote: On 12/19/2013 1:06 PM, Joseph Kregloh wrote: It's easier to keep things segregated. It is not anymore different than doing the upgrade in the same jail. Which at the end of the day you are doing the upgrade in the same jail, because at the end of the day pg_upgrade just needs the old data an binary to start and create some dump files. pg_upgrade needs to access the old data AND all the tablespaces at the same paths as the old server sees them AND the new data and tablespaces at the same path as the NEW server sees them. if the two servers are in different jails, I don't see how you could make that work... if you run pg_upgrade in the host system, then all the paths are different for both sets of data and tablespaces. I understand that it will need to access the old data and new data data as it sees it, but it is seeing everything as /usr/local/pgsql/data. Now lets say I have both versions 9.0 and 9.3 installed in the same jail. They will both need to use /usr/local/pgsql/data to access the physical data. But that will not work because all of the Postgres related files are in there, so you can only have 9.0 OR 9.3 use the /usr/local/pgsql/data directory.
No, that is not the case. The data directory can be different for different instances, it is a configure option. In fact the pg_upgrade docs point that out:
http://www.postgresql.org/docs/9.3/interactive/pgupgrade.html See: Usage Steps 1-3
-- Adrian Klaver adrian.klaver@xxxxxxxxx -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
