Horton

Re: Add jsonlog log_destination for JSON server logs

Michael Paquier <michael@paquier.xyz>

View thread
From: Michael Paquier <michael@paquier.xyz>
To: Sehrope Sarkuni <sehrope@jackdb.com>
Cc: Pg Hackers <pgsql-hackers@postgresql.org>, david@fetter.org
Date: 2021-09-14T06:06:41Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Introduce log_destination=jsonlog

  2. Refactor set of routines specific to elog.c

  3. Fix compilation warning in syslogger.c

  4. Refactor per-destination file rotation in logging collector

  5. Refactor output file handling when forking syslogger under EXEC_BACKEND

  6. Add regression tests for csvlog with the logging collector

  7. Refactor the syslogger pipe protocol to use a bitmask for its options

On Mon, Sep 13, 2021 at 11:56:52PM -0400, Sehrope Sarkuni wrote:
> Good catch. Staring at that piece again, that's tricky as it tries to
> aggressively free the buffer before calling write_cvslog(...). Which can't
> just be duplicated for additional destinations.
> 
> I think we need to pull up the negative case (i.e. syslogger not available)
> before the other destinations and if it matches, free and exit early.
> Otherwise, free the buffer and call whatever destination routines are
> enabled.

Yes, I got a similar impression.

> I started on a bit of this as well. There's so much overlap already between
> the syslog_ and csvlog code that I'm going to put that together first. Then
> the json addition should just fall into a couple of call sites.
> 
> I'm thinking of adding an internal struct to house the FILE* and file
> names. Then all the opening, closing, and log rotation code can pass
> pointers to those and centralize the pfree() and NULL checks.

Agreed on both points (using a structure and doing the refactoring as
a first patch).
--
Michael