On 16/12/10 12:12, Eric Comeau wrote:
> The problem surfaced at one of our client installs where a maintenance
> DELETE query was running for over 24 hrs. We have since then identified
> the missing indexes and have sent the client a script to create them,
> but in our testing we could not been able to get postgres to use the new
> index for the FK cascade delete without bouncing the database.
Well, an ongoing DELETE isn't going to see a new index. I'd have thought
a new connection should though.
In response to
pgsql-performance by date
|Next:||From: Eric Comeau||Date: 2010-12-16 12:55:46|
|Subject: Re: How to get FK to use new index without restarting the
|Previous:||From: Richard Huxton||Date: 2010-12-16 12:37:50|
|Subject: Re: performance libpq vs JDBC|