PostgreSQL 9 'pgstats wait timeout' problem

Question

I'm asking about this here because in the PostgreSQL mailing lists no one answered me. In one of my servers (Windows 2008 R2 Standard/PostgreSQL 9.0.5 64 bits) the logs are being flooded with the following message:

2012-03-26 00:00:11 BRT AVISO:  pgstat wait timeout
2012-03-26 00:00:19 BRT AVISO:  pgstat wait timeout
2012-03-26 00:00:31 BRT AVISO:  pgstat wait timeout

I checked other threads in the PostgreSQL mailing lists that were answered, but there are no definitive answer. Some speculate about I/O overload, but I don't think it would cause this message to be logged almost all the time that the server is up. Do you, that use the combination of Windows+Postgresql, have ever had this problem? If yes, were you able to fix or identify the problem?

Thanks

Answer 1

Reloading the config makes the errors go away, and performance improves (and db statistics start getting recorded again).

I've set up a Windows Scheduled Task which performs a pg_ctl reload -D"insert here the address of your pg directory" on the hour (we're running an incredibly busy application). Hacky, but it's working.