Thanx, I see the difference now. The first entry will always start at an epoch time that's divisible by the stepsize. I thought it would start at the time of the first update.
Taking 3600 as the stepsize is logical since the localtimes change with 1 hour steps. You were right about the xff remark too, I've learned today :) Serge. -----Original Message----- From: Alex van den Bogaerdt [mailto:[EMAIL PROTECTED] Sent: Friday, April 19, 2002 11:23 PM To: Serge Maandag Cc: RRD users Subject: Re: [rrd-users] Re: RRD setup when collection only daily Serge Maandag wrote: > If he sets his step to 86400 and his ds heartbeat to: > 86400 + (a small interval in case he runs late with his daily update) > And he does his first update at midnight, wouldn't that just be what > he wants? His pdp's would be at 00:00 exactly. No, unless you work in UTC time. All times inside RRDtool are a whole multiple of the step size. If the step size is 86400, today would be 1019174400. This is 1019250638 - (1019250638 % 86400). Perl shows the time: perl -e 'use POSIX;print ctime(1019174400)' which results in 'Fri Apr 19 02:00:00 2002' (local time). > I can see why your method works, but wonder what the advantage is. If the step size is 3600, the current time minus the modulo would be perl -e 'use POSIX;print ctime(1019250638 - (1019250638 % 3600));' Fri Apr 19 23:00:00 2002 So, if I would update now, in the first example I would update at 02:00 local time, in the second example I would update at 23:00 local time. Similarly, if I would update just after midnight I would still update at 02:00 (the day before!) in the first example while I would update at 00:00 (*that* day) in the second example. -- Unsubscribe mailto:[EMAIL PROTECTED] Help mailto:[EMAIL PROTECTED] Archive http://www.ee.ethz.ch/~slist/rrd-users WebAdmin http://www.ee.ethz.ch/~slist/lsg2.cgi
