Re: Vacuum process waiting on BufferPin

On 14/08/18 15:26, Don Seiler wrote:
> On Tue, Aug 14, 2018 at 8:23 AM, Vick Khera <vivek@khera.org
> <mailto:vivek@khera.org>> wrote:
> 
> 
>     The general fix is to never sit idle in transaction, but this
>     specific case closing the cursor seems like it will also do it.
> 
> 
> Of course. I've let development know that they need to sort out why it's
> left that way.

There is an idle_in_transaction_session_timeout parameter to kill
connections that are idle (in transaction) for too long.  It was
implemented specifically for cases like this.
-- 
Vik Fearing                                          +33 6 46 75 15 36
http://2ndQuadrant.fr     PostgreSQL : Expertise, Formation et Support