Thread: Statistics collector failure messages on startup

Statistics collector failure messages on startup

From
Christophe Pettus
Date:
On PostgreSQL 9.0.1, on Ubuntu 10.04.2, I'm seeing these on startup:

2011-09-22 19:22:15.984 UTC,,,6066,,4e7b8ae7.17b2,1,,2011-09-22 19:22:15 UTC,,0,LOG,08006,"test message did not get
throughon socket for statistics collector",,,,,,,,,"" 
2011-09-22 19:22:15.984 UTC,,,6066,,4e7b8ae7.17b2,2,,2011-09-22 19:22:15 UTC,,0,LOG,00000,"trying another address for
thestatistics collector",,,,,,,,,"" 

... along with "WARNING:  pgstat wait timeout" when I attempt to use any stats-collector-related info.  I've confirmed
thatthe loopback interface is unfiltered for UDP.  The stats collector process itself is up and running, and the temp
fileis writeable. 

Any thoughts?

--
-- Christophe Pettus
  xof@thebuild.com


Re: Statistics collector failure messages on startup

From
Tom Lane
Date:
Christophe Pettus <xof@thebuild.com> writes:
> On PostgreSQL 9.0.1, on Ubuntu 10.04.2, I'm seeing these on startup:
> 2011-09-22 19:22:15.984 UTC,,,6066,,4e7b8ae7.17b2,1,,2011-09-22 19:22:15 UTC,,0,LOG,08006,"test message did not get
throughon socket for statistics collector",,,,,,,,,"" 
> 2011-09-22 19:22:15.984 UTC,,,6066,,4e7b8ae7.17b2,2,,2011-09-22 19:22:15 UTC,,0,LOG,00000,"trying another address for
thestatistics collector",,,,,,,,,"" 

> ... along with "WARNING:  pgstat wait timeout" when I attempt to use
> any stats-collector-related info.  I've confirmed that the loopback
> interface is unfiltered for UDP.

I'd say better check that again.  The test that's whining was designed
specifically to detect cases where a packet filter prevents pg_stats
traffic from getting through.  Frequently this is an IPv4 vs IPv6 kind
of problem...

            regards, tom lane