Hi Clayton,

the failover time is steered by the time_reopen() global option (60 sec by default), so setting it to a much lower value should do the trick.

From the admin guide:
"To list several failover servers, separate the address of the servers with comma. The time syslog-ng PE waits for the a server before switching to the next failover server is set in the time_reopen() option. For details about how client-side failover works, see Section 2.15, Client-side failover."

http://www.balabit.com/sites/default/files/documents/syslog-ng-pe-v4.0-guide-admin-en.html/index.html-single.html#example-failover-server-tcp

I assume that the TCP connection breakdown is detected by syslog-ng in a timely manner, pls correct me if I misunderstood the problem.

Balint

On 10/04/2011 12:18 PM, Clayton Dukes wrote:

Hi Folks,
We have a team doing testing on the PE edition and they are finding that the timeout for the failover trigger takes 60 seconds:

It was observed that it takes anytime between 10 to 60 seconds to detect connection is broken. We found out it by running syslog-ng in debug mode. I guess it relies on connection timeout. This 60 second is large interval compared to the default interval that we have (15 sec).

My question: Is there a way to tune this parameter down to 15 seconds in the config?

______________________________________________________________

Clayton Dukes
______________________________________________________________
______________________________________________________________________________
Member info: https://lists.balabit.hu/mailman/listinfo/syslog-ng
Documentation: http://www.balabit.com/support/documentation/?product=syslog-ng
FAQ: http://www.balabit.com/wiki/syslog-ng-faq