<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN"> <HTML> <HEAD> <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1"> <META NAME="Generator" CONTENT="MS Exchange Server version 5.5.2658.24"> <TITLE>FW: Problem with initial service scheduling (2.0b3)</TITLE> </HEAD> <BODY> Sorry for posting this message again but I cannot modify my production environment before having an opinion from somebody who understands the 'interleave_block' stuff. Thanks in advance > -----Original Message----- > From: Laupretre, François (CALYON) > Sent: Thursday, June 09, 2005 2:56 PM > To: nagios-devel@lists.sourceforge.net > Subject: Problem with initial service scheduling (2.0b3) > > > Hi all, > > I currently have a configuration with 4800 services : 600 > active and 4200 passive. And, as the number was growing, I > noticed a problem in the way nagios scheduled their initial > check time : With the 2.0b3 original code, with > max_service_check_spread=30, when I look at the scheduling > queue just after start, I see that the last service checks > are scheduled to run in 4 hours ! > > This delay corresponds to : > > max_service_check_spread * (total_services / total_scheduled_services) > > And should be equal to max_service_check_spread. > > I found the reason in event.c/init_timing_loop() and I am > including a change which appears to correct the problem but, > as I am not sure to fully understand the 'interleave_block' > logic, this change should be taken with care : > > The reason : in the 'schedule service checks' section of > init_timing_loop(), next check time is incremented for each > service, and not for each SCHEDULED service. So, in my case > it is incremented 'total_services' times and the last check > time is equal to : > > Current_time + total_services * service_inter_check_delay > > Where it should be : > > Current_time + total_scheduled_services * service_inter_check_delay > > Which is coherent with the way service_inter_check_delay is computed. > > My change consists of taking the 'should_be_scheduled' check > out of the inner loop, and add a line in order to have the > code enter the inner 'interleave_block' loop only for active > checks. This way current_interleave_block goes from 0 to > total_schedules_services instead of going up to total_services. > > Once again, the patch I am submitting seems to correct the > problem in MY case. But I don't know if it is correct when > interleave variables have some different values. > > Regards > > François > </BODY> </HTML>