Обсуждение: syslog configurable line splitting behavior

Writing log messages to syslog caters to ancient syslog implementations
in two ways:

- sequence numbers
- line splitting

While these are arguably reasonable defaults, I would like a way to turn
them off, because they get in the way of doing more interesting things
with syslog (e.g., logging somewhere that is not just a text file).

So I propose the two attached patches that introduce new configuration
Boolean parameters syslog_sequence_numbers and syslog_split_lines that
can toggle these behaviors.

On 02/27/2016 04:49 AM, Peter Eisentraut wrote:
> So I propose the two attached patches that introduce new configuration
> Boolean parameters syslog_sequence_numbers and syslog_split_lines that
> can toggle these behaviors.

As someone who uses syslog for my servers I find both of these GUCs 
useful, especially when used in combination, and I do not think a 
compile time option like suggest by Alexander would be suitable 
substitute because then I would need a custom build of PostgreSQL just 
to change this which seems too much effort just for this.

The code itself is clean and I find the documentation to be easy to 
understand.

I have one nitpick: why is one of the variables "true" while the other 
is "on" in the example? I think both should be "on".

#syslog_sequence_numbers = true
#syslog_split_lines = on

Another possible improvement would be to change "Split messages sent to 
syslog." to something more verbose like "Split messages sent to syslog, 
by lines and to fit in 1024 bytes.".

I tested the code and it worked as advertised and also passed the test 
suite. My local rsyslogd cut the message at about 8 kB, but I do not 
think it is worth it to expose PG_SYSLOG_LIMIT as a GUC.

Andreas

On 3/4/16 11:01 AM, Alexander Korotkov wrote:
> On Sat, Feb 27, 2016 at 6:49 AM, Peter Eisentraut <peter_e@gmx.net
> <mailto:peter_e@gmx.net>> wrote:
> 
>     Writing log messages to syslog caters to ancient syslog implementations
>     in two ways:
> 
>     - sequence numbers
>     - line splitting
> 
>     While these are arguably reasonable defaults, I would like a way to turn
>     them off, because they get in the way of doing more interesting things
>     with syslog (e.g., logging somewhere that is not just a text file).
> 
>     So I propose the two attached patches that introduce new configuration
>     Boolean parameters syslog_sequence_numbers and syslog_split_lines that
>     can toggle these behaviors.
> 
> 
> Would it have any usage if we make PG_SYSLOG_LIMIT configurable (-1 for
> disable) instead of introducing boolean?

That would work, too.  But then we'd need another setting to disable
splitting on newlines.  That way we'd have more settings, but they
actually mirror the corresponding settings on the rsyslogd side better.

On 3/8/16 9:12 PM, Andreas Karlsson wrote:
> As someone who uses syslog for my servers I find both of these GUCs
> useful, especially when used in combination, and I do not think a
> compile time option like suggest by Alexander would be suitable
> substitute because then I would need a custom build of PostgreSQL just
> to change this which seems too much effort just for this.

I think he was suggesting to take the existing compile-time constant and
make a run-time setting out of it.

On 03/09/2016 03:20 AM, Peter Eisentraut wrote:
> On 3/8/16 9:12 PM, Andreas Karlsson wrote:
>> As someone who uses syslog for my servers I find both of these GUCs
>> useful, especially when used in combination, and I do not think a
>> compile time option like suggest by Alexander would be suitable
>> substitute because then I would need a custom build of PostgreSQL just
>> to change this which seems too much effort just for this.
>
> I think he was suggesting to take the existing compile-time constant and
> make a run-time setting out of it.

Ah, that makes more sense than my reading of his message. Still I do not 
think there would be much gain from doing so. I do not see much benefit 
from making it configurable. In the cases I envision you would either 
want to split with a log file in mind or not split at all.

Andreas

On 3/8/16 9:12 PM, Andreas Karlsson wrote:
> I have one nitpick: why is one of the variables "true" while the other
> is "on" in the example? I think both should be "on".
>
> #syslog_sequence_numbers = true
> #syslog_split_lines = on
>
> Another possible improvement would be to change "Split messages sent to
> syslog." to something more verbose like "Split messages sent to syslog,
> by lines and to fit in 1024 bytes.".

Updated patches with your suggestions.  I also renamed
syslog_split_lines to syslog_split_messages, which I think is more accurate.

On 03/16/2016 03:50 AM, Peter Eisentraut wrote:
> On 3/8/16 9:12 PM, Andreas Karlsson wrote:
>> I have one nitpick: why is one of the variables "true" while the other
>> is "on" in the example? I think both should be "on".
>>
>> #syslog_sequence_numbers = true
>> #syslog_split_lines = on
>>
>> Another possible improvement would be to change "Split messages sent to
>> syslog." to something more verbose like "Split messages sent to syslog,
>> by lines and to fit in 1024 bytes.".
>
> Updated patches with your suggestions.  I also renamed
> syslog_split_lines to syslog_split_messages, which I think is more accurate.

I think the patch looks good now. I will change the status to "Ready for 
committer".

Andreas