Re: BUG #7890: wrong behaviour using pg_rotate_logfile() with parameter log_truncate_on_rotation = on

On Monday, February 18, 2013 5:42 PM sari thiele wrote:
> The following bug has been logged on the website:
>
> Bug reference: 7890
> Logged by: Sari Thiele
> Email address: sari(dot)thiele(at)usit(dot)uio(dot)no
> PostgreSQL version: 9.1.8
> Operating system: Red Hat Enterprise Linux Server release 6.3
> Description:
>
> Hi,
>
> I am trying to use the function pg_rotate_logfile() to force a truncate
> of
> the current log file by setting the parameter "log_truncate_on_rotation
> =
> on".
> But it looks like that this does not work, like I expected.
>
> System info:
> Red Hat Enterprise Linux Server release 6.3 (Santiago)
>
> select version();
> version
>
> -----------------------------------------------------------------------
> ---------------------------------------
> PostgreSQL 9.1.8 on x86_64-unknown-linux-gnu, compiled by gcc (GCC)
> 4.4.6
> 20120305 (Red Hat 4.4.6-4), 64-bit
>
>
> Parameter set-up:
>
> log_filename = 'postgresql-%F_%H.log'
> logging_collector = on
> log_truncate_on_rotation = on
> log_rotation_age = 1h
> log_rotation_size = 0
>
> SELECT pg_rotate_logfile();
>
> After the function call, the file is still filled with the previous
> data.
> After my understanding the log file should be emptied (truncated).
>
> -----------
> To check if there is something wrong with my set-up, I did some other
> tests,
> too.
>
> 1.
> log_filename = 'postgresql-%F_%T.log'
> logging_collector = on
> log_truncate_on_rotation = on
> log_rotation_age = 1h
> log_rotation_size = 0
>
> SELECT pg_rotate_logfile();
>
> After the function call a new file is generated, because of the
> logfilename
> string (HH:MM:ss).
> This is correct.
>
> -------------------------
> 2.
> log_filename = 'postgresql-%M.log'
> logging_collector = on
> log_truncate_on_rotation = on
> log_rotation_age = 10min
> log_rotation_size = 0
>
> Here is the file automatically truncated after an hour, because of the
> file
> name. So that works also correct.
>
> ----
>
> We checked the source code and it looks like the problem could be here:
>
> src/backend/postmaster/syslogger.c
>
>
> /*
> * Decide whether to overwrite or append. We can overwrite if (a)
> * Log_truncate_on_rotation is set, (b) the rotation was triggered by
> * elapsed time and not something else, and (c) the computed file name
> is
> * different from what we were previously logging into.
> *
> * Note: last_file_name should never be NULL here, but if it is,
> append.
> */
> if (time_based_rotation || (size_rotation_for
> &LOG_DESTINATION_STDERR))
> {
> if (Log_truncate_on_rotation && time_based_rotation &&
> last_file_name != NULL &&
> strcmp(filename, last_file_name) != 0) ==> *
> fh = logfile_open(filename, "w", true);
> else
> fh = logfile_open(filename, "a", true);
>
>
> ==> * this checks if the new file name is different from the old one.
> In my case it is still the same name and so the new log lines will be
> appended to the current file, not truncated like I expected.
> I think this is the wrong behaviour, what do you think?

The only reason is you have used pg_rotate_logfile() to perform the action, otherwise if it would have been done
based on time, it should have truncated the file.
Why you think this is wrong behavior, do you expect any time you call pg_rotate_logfile(), it should truncate the file if
log_truncate_on_rotation is on?
I think if you are expecting such behavior, it might not be right expectation, because it considers the the time and log_filename format as well.

With Regards,
Amit Kapila.