[rsyslog] PostgreSQL: Problems with character encoding

[Rainer Gerhards]

> -----Original Message-----
> From: rsyslog-bounces at lists.adiscon.com [mailto:rsyslog-
> bounces at lists.adiscon.com] On Behalf Of Jakob Haufe
> Sent: Sunday, January 24, 2010 8:15 PM
> To: rsyslog at lists.adiscon.com
> Subject: Re: [rsyslog] PostgreSQL: Problems with character encoding
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Fri, 22 Jan 2010 10:51:41 +0100
> "Rainer Gerhards" <rgerhards at hq.adiscon.com> wrote:
> 
> > V5 has the capability to discard messages that cause an action
> failure.
> > However, this is mostly untested yet, AND the action must support it
> by
> > providing proper status information - it must differentiate between
> > system-induced errors (which can be retried) and message-induced
> errors
> > (which need the discard). ompgsql currently does not provide that
> status
> > information.
> 
> If you can point me at some example code or docs on how to do this, I
> would
> like to try and add this functionality to ompgsql. Does ommysql already
> implement that?


It's pretty new functionality and there is not yet a good example plugin that
uses it (it makes most sense for database plugins, where I have limited
knowledge). It would be useful to read this first (unfortunately not an easy
read:

http://download.rsyslog.com/design.pdf

Actually implementing it is rather easy. The core point is that for
system-induced errors (those that can be retried) the plugin must return
RS_RET_SUSPENDED and for message-induced errors it must return an "real"
error state (like RS_RET_ERR, but it would be better, and I'd be glad to
include, more precise error codes). The core engine than knows what to do.
Well, the core may have undiscovered bugs right now, as this functionality
was never before used in practice. It is very criticial to think about which
error class a failure belongs to. Messages with message-induced errors are
simply thrown away, so one needs to think twice before assigning this class -
but on the contrary if such a message is flagged as system-induced, it will
block the system, just as you can currently see...

HTH
Rainer

> 
> Regards,
> Jakob Haufe (sur5r)
> 
> - --
> ceterum censeo microsoftem esse delendam.
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.10 (GNU/Linux)
> 
> iEYEARECAAYFAktcnCEACgkQ1YAhDic+ada9WACeMkawcNTL/lt5E70mWeVjd38G
> ARoAn1OAkEqm7NXRMwwVzUDC3B/2TeCB
> =eDPw
> -----END PGP SIGNATURE-----
> _______________________________________________
> rsyslog mailing list
> http://lists.adiscon.net/mailman/listinfo/rsyslog
> http://www.rsyslog.com