Обсуждение: stats collector appears to be dying

I'm having an issue with what appears to be the stats
collector process dying on Postgresql 8.0.1 running on
Windows XP w/sp1.

I've enabled
stats_command_string and stats_row_level in my config
file.  During bulk inserts the stats collector process
appears to encounter some sort of problem and shuts
down.   The end result is the log file being spammed
with "FATAL" entries regarding the process.  I've
attached a small excerpt from my log files.

Any ideas as to what is going on?

Regards,

Shelby Cain



__________________________________
Do you Yahoo!?
Yahoo! Mail - Helps protect you from nasty viruses.
http://promotions.yahoo.com/new_mail

Windows firewall perhaps?

Shelby Cain wrote:

>I'm having an issue with what appears to be the stats
>collector process dying on Postgresql 8.0.1 running on
>Windows XP w/sp1.
>
>I've enabled
>stats_command_string and stats_row_level in my config
>file.  During bulk inserts the stats collector process
>appears to encounter some sort of problem and shuts
>down.   The end result is the log file being spammed
>with "FATAL" entries regarding the process.  I've
>attached a small excerpt from my log files.
>
>Any ideas as to what is going on?
>
>Regards,
>
>Shelby Cain
>
>
>
>__________________________________
>Do you Yahoo!?
>Yahoo! Mail - Helps protect you from nasty viruses.
>http://promotions.yahoo.com/new_mail
>
>------------------------------------------------------------------------
>
>
>---------------------------(end of broadcast)---------------------------
>TIP 5: Have you checked our extensive FAQ?
>
>               http://www.postgresql.org/docs/faq
>
>

--- "Matthew T. O'Connor" <matthew@zeut.net> wrote:

> Windows firewall perhaps?
>

I double checked to make sure it was disabled.
Regardless, if a firewall were preventing the
connection from occuring I'm fairly certain the error
would be something other than "target machine actively
refused" which implies trying to connect() to a port
that is open but doesn't have a listen socket bound to
it.

Anyway, the first line in my log mentioning the stats
collector process is the following:

"2005-02-23 14:59:17 FATAL:  could not read from
statistics collector pipe: No such file or directory"

I suspect that is the critical clue as to what might
be happening on this box.    This is a default
installation with a few deviations from the supplied
default config.  Any other ideas?

Regards,

Shelby Cain

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

Shelby Cain <alyandon@yahoo.com> writes:
> I'm having an issue with what appears to be the stats
> collector process dying on Postgresql 8.0.1 running on
> Windows XP w/sp1.

> 2005-02-23 14:59:17 FATAL:  could not read from statistics collector pipe: No such file or directory
> 2005-02-23 14:59:17 LOG:  statistics collector process (PID 2748) was terminated by signal 1

Odd.  We probably shouldn't take the message entirely at face value
because of the error code mapping that happens under Windows.
Looking at backend/port/win32/error.c, there seem to be several Windows
codes that are translated to ENOENT:

ERROR_FILE_NOT_FOUND
ERROR_PATH_NOT_FOUND
ERROR_INVALID_DRIVE
ERROR_NO_MORE_FILES
ERROR_BAD_NETPATH
ERROR_BAD_NET_NAME
ERROR_BAD_PATHNAME
ERROR_FILENAME_EXCED_RANGE

I wonder if any of those throws more light on it ...

            regards, tom lane

> I'm having an issue with what appears to be the stats
> collector process dying on Postgresql 8.0.1 running on
> Windows XP w/sp1.
>
> I've enabled
> stats_command_string and stats_row_level in my config file.
> During bulk inserts the stats collector process appears to
> encounter some sort of problem and shuts
> down.   The end result is the log file being spammed
> with "FATAL" entries regarding the process.  I've attached a
> small excerpt from my log files.
>
> Any ideas as to what is going on?

This is a known problem on WIndows. THe stats processor sometimes falls
down under heavly load. And if you enable row-level-stats the load goes
up considerably - especially during bulk inserts.

//Magnus

--- Magnus Hagander <mha@sollentuna.net> wrote:

>
> This is a known problem on WIndows. THe stats
> processor sometimes falls
> down under heavly load. And if you enable
> row-level-stats the load goes
> up considerably - especially during bulk inserts.
>
> //Magnus
>

Well then I guess I'll be leaving that off for the
time being.  Thanks for the response.  :)

Regards,

Shelby Cain



__________________________________
Do you Yahoo!?
Yahoo! Mail - Easier than ever with enhanced search. Learn more.
http://info.mail.yahoo.com/mail_250