Re: [Sqlgrey-users] 1.3.2-1 re-try, not quite yet

[Lionel B. <lio...@bo...>]

HaJo Schatz wrote the following on 11/18/04 18:39 :

>I know I'm getting annoying... I just tried and brought PostgreSQL down
>while sqlgrey was running. The logs said:
>
>Nov 19 01:16:08 sun sqlgrey[32194]: Warning: couldn't do query: DELETE
>FROM from_awl WHERE last_seen < now() - INTERVAL '60 DAY': FATAL:  This
>connection has been terminated by the administrator., reconnecting to DB
>Nov 19 01:16:08 sun sqlgrey[32194]: Can't connect to DB: could not
>connect to server: Connection refused
>Nov 19 01:16:08 sun sqlgrey[32194]: warning: Database handle destroyed
>without explicit disconnect at /usr/bin/sqlgrey line 608.
>Nov 19 01:16:08 sun sqlgrey[32194]: fatal: Can't call method "do" on an
>undefined value at /usr/bin/sqlgrey line 77.
>
>And sqlgrey disappeared from my process list...
>  
>

Ok, so my tests weren't enough : I dropped each table one at a time to 
check what would happen in different parts of the code when a database 
error occured. What I didn't expect is that a lost connection could have 
different effects than a pure SQL error. But now it makes sense that if 
the database isn't available, the database handle object is destroyed 
and calling the do method on it isn't such a great idea :-(

Back to the drawing board (fault tolerance isn't a piece of cake, if we 
can't rely on the dbh object, I'll have to wrap it in another object).

Thanks

Lionel : still learning DBI :-)