On Mon, 3 Jun 2002, Dieter Simader wrote: > On Mon, 3 Jun 2002, Jonas Smedegaard wrote:
> > Dieter: I imagine the source of the problem lies in using a more > > strict access to the postgres server than the average setup (never > > allowing master access from the web browser, only a specific account > > holding all ledger databases, and the use the suexec feature of Apache > > to run as that same user). > The problem is with not using the right database user. If you create a > dataset with the name postgres and you enter pgsql for the upgrade there > is nothing to upgrade for the user. As I remember it I created the databases with the user postgres (on the commandline from a backup - earlier problems with Postgres upgrade required manual editing) and then allowed another user (the one Apache runs as) fulle access to those databases. > The proper procedure to upgrade is to check the database name you use by > clicking on one of the logins. Then you use this database name for the > upgrade. Which is what I did with no luck! Using the Postgres login and password used for normal SQL-Ledger access resulted in an "All databases up to date" screen (which in fact was "no databases found to update"). Removing the owner check (or is that what I did?) as described in the diff solved the problem for me: I could then use "the database name you use by clicking on one of the logins" exactly is before - only now I was presented with a batch of databases to upgrade. > > Beware that this can ruin non-postgres databases hosted on the same > > postgres server. But then again - you get a second screen where you > > can unselect wrongly selected databases by hand before you do the > > actual update (and you DID use separate login for each set of > > databases and avoid master login from a browser, right?). > I should also point out that, contrary to Jonas' believe, you cannot > ruin anything because not only do a I check if the dataset belongs to > the user but also check if the dataset is in fact one which belongs to > SQL-Ledger. My warning was regarding _hacked_ scripts, not your original one. I didn't investigate further, just quickly spotted that what I did lowered your checks. Sorry if I didn't make that clear. - Jonas -- Jonas Smedegaard (+45 40843136) http://dr.jones.dk/~jonas/ Spiff ApS (= IT-guide dr. Jones ApS) http://dr.jones.dk/ Debian GNU/Linux developer http://people.debian.org/~js/ _______________________________________________________________ Don't miss the 2002 Sprint PCS Application Developer's Conference August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm _______________________________________________ Listinfo https://lists.sourceforge.net/lists/listinfo/sql-ledger-users Archive http://www.geocrawler.com/redir-sf.php3?list=sql-ledger-users
