Re: [GENERAL] Efficiently Triggering Autovacuum Analyze?

2015-12-31 Thread Jim Nasby
On 12/30/15 1:31 PM, Joe Conway wrote: On 12/30/2015 11:09 AM, Cory Tucker wrote: With this scenario you can expect an autoanalyze every 5 million rows and autovacuum every 10 million. In my experience (and based on your description, yours as well) this is not often enough. Not only that, when

Re: [GENERAL] Efficiently Triggering Autovacuum Analyze?

2015-12-30 Thread Tom Lane
Cory Tucker writes: > This table is almost always queried using a combination of (account_id, > record_id) and is generally pretty fast. However, under certain loads, the > query becomes slower and slower as time goes on. The workload that causes > this to happen is when

Re: [GENERAL] Efficiently Triggering Autovacuum Analyze?

2015-12-30 Thread Cory Tucker
On Wed, Dec 30, 2015 at 11:20 AM Tom Lane wrote: > Cory Tucker writes: > > This table is almost always queried using a combination of (account_id, > > record_id) and is generally pretty fast. However, under certain loads, > the > > query becomes

[GENERAL] Efficiently Triggering Autovacuum Analyze?

2015-12-30 Thread Cory Tucker
We have a performance problem accessing one of our tables, I think because the statistics are out of date. The table is fairly large, on the order of 100M rows or so. The general structure of the table is as follows: Column | Type | Modifiers

Re: [GENERAL] Efficiently Triggering Autovacuum Analyze?

2015-12-30 Thread Joe Conway
On 12/30/2015 11:09 AM, Cory Tucker wrote: > We have a performance problem accessing one of our tables, I think > because the statistics are out of date. The table is fairly large, on > the order of 100M rows or so. > The fix I have employed to restore the speed of the query after I notice > it