Thread: keeping a log / debug info

keeping a log / debug info

From
"Johnson, Shaunn"
Date:

Howdy:

Running PostgreSQL 7.2.1 on RedHat Linux 7.2 kernel 2.4.7.

I am running postgres with a '-d' option (debug)
and storing that info in a log file.  This morning
I moved the log file (trying to gzip it) and touched
a new version of that file ... (didn't mean to do that).
Now no data is being recorded ...

Can I restart the database without losing backends
that currently exist?  Sort of a -HUP postmaster<pid>?

Is there a way to record this info via /etc/syslog.conf
and have the system do it? (for instance, the message file ...
when it gets so large, it is moved to message.0
and you see a *new* message file).

Thanks!

-X

Re: keeping a log / debug info

From
Tom Lane
Date:
"Johnson, Shaunn" <SJohnson6@bcbsm.com> writes:
> I am running postgres with a '-d' option (debug)
> and storing that info in a log file.  This morning
> I moved the log file (trying to gzip it) and touched
> a new version of that file ... (didn't mean to do that).
> Now no data is being recorded ...

If you are just directing stderr into a file then you can't rotate
log files that way.  You could use syslog instead, or you could
pipe stderr to a log-rotation script.  I prefer the latter, mainly
because there are some messages that can't be redirected to syslog
(eg, dynamic linker failure messages).  See
http://www.ca.postgresql.org/users-lounge/docs/7.2/postgres/logfile-maintenance.html
and past discussions in the mailing list archives.

            regards, tom lane

Re: keeping a log / debug info

From
"Shridhar Daithankar"
Date:
On 19 Sep 2002 at 11:53, Johnson, Shaunn wrote:

>
> Howdy:
> Running PostgreSQL 7.2.1 on RedHat Linux 7.2 kernel 2.4.7.
> I am running postgres with a '-d' option (debug)
> and storing that info in a log file. This morning
> I moved the log file (trying to gzip it) and touched
> a new version of that file ... (didn't mean to do that).
> Now no data is being recorded ...
> Can I restart the database without losing backends
> that currently exist? Sort of a -HUP postmaster<pid>?
> Is there a way to record this info via /etc/syslog.conf
> and have the system do it? (for instance, the message file ...
> when it gets so large, it is moved to message.0
> and you see a *new* message file).

I can think of way to do this. Taken from apache. It can pipe the logs to
another process and compress compress it etc...

Check apache/logs.html#rotation in apache documentation.. May be you can pipe
the logs to your own program which roatates the log or may be logrotate
itself..

HTH..

Bye
 Shridhar

--
weapon, n.:    An index of the lack of development of a culture.