Hi,
As of now, i found the following cases where we can expect these kind of
WARNING message in pg_log.
Case 1 { Huge I/O }
======
When the postgresql autovacuum process is not able to get the required I/O
to write the statistics to "stats_temp_location" then we can get this kind
of WARNING Message. As discussed, the huge I/O may causing due to the
checkpoint occurs on the Database.
How to log the checkpoints information .
===============================
1) Edit the postgresql.conf file as log_checkpoints=on
2) Select Pg_Reload_Conf();
Case 2 {Invalid stats_temp_location}
======
When the postgresql "stats_temp_location" is invalid path, then in this
case also we can expect this kind of WARNING Message. If you want to change
this location to some other place, then we need to follow the below
approach.
1) Edit the postgresql.conf file as stats_temp_location='<PATH>'
2) Select Pg_Reload_Conf();
Case 3 {Invalid Localhost IP}
======
There might be a chance, we have an invalid Localhost IP. Please check the
localhost entires in Hosts file and rectify it if any thing wrong.
Once we made any changes in this file then we need to restart the
PostgreSQL Cluster to take it's Effect on autovacuum worker processes.
I hope some one will add more on this.
Regards,
Dinesh Kumar
manojadinesh.blogspot.com
On Mon, Feb 4, 2013 at 5:24 PM, Jake Stride <jake@stride.me.uk> wrote:
> I have had some issues with a database on EC2 and I have restored it to a
> new instance. When vacuuming the database I am getting the following in the
> logs;
>
> WARNING: pgstat wait timeout
>
> Is this normal/acceptable?
>
> Thanks
>