On Fri, 2008-08-15 at 17:19 +0300, Kostyantyn Gushtin wrote:
Hello,
there is an encoding problem if using syslog-ng with windows agent.
Logs from Linux hosts are written to file as UTF-8, while the logs from windows agent are written to file as cp1252. The problem is visible if you have a message with chars out of ASCII set, like Ö. As a result we have a file that is both unreadable and useless for automation.
Is there any way to fix this? Or do you plan to fix this?
From my point of view syslog-ng should write file using one charset. Preferably configurable.
syslog-ng 2.1 (just as other syslogds) don't care about character sets, it uses anything it receives. On the other hand the forthcoming syslog-ng 3.0 has support for character sets, but that's not yet released. (it is able to convert input to UTF8) The 3.0 version of the Windows agent converts everything to UTF8. The Windows agent has already been released, but not really publicized as we're waiting for both of them to be released at the same time. You can get your copy if you write an email to Zoltan or Eniko. On the OSE front, I plan to backport a lot of features to 3.0 OSE this time, I just have to find the proper timing. (to avoid having to synchronize the two branches a lot), but more on this later if it really becomes available. -- Bazsi