[rrd-users] Re: Problems with multi-DS RRD that gets asynchronous-DS updates

[steve rader]

 > > From: me
 > > [...]
 > > Then at 1999-09-16 12:45:00 CDT, it did snmpgets across
 > > a second boundary so my script stuffed values into rrd via:
 > > 
 > >   rrdtool update MadisonSD.rrd -t input 937503900:496053104
 > >   rrdtool update MadisonSD.rrd -t output 937503901:108120595
 > > 
 > > I expect the value of the DS output at 1999-09-16 12:45:00 CDT to be
 > > 1.058e+04 = (108120595 - 104956107) / 300.
 > > 
 > > But instead I get NaN:
 > > [...]
  
 > From: Alex van den Bogaerdt
 > Been thinking about it for a while. What could happen is that
 > rrd writes an unknown at time=937503900, verify by dumping the database
 > inbetween 937503900 and 937503901.

Yep.  That's correct. 

 > Perhaps updating at 937503899 and 937503900 will do the trick ?

Nope.  I'm quite sure an update of the DS input at 937503899
will cause the DS output to be unknown (U) (like above).
(And besides, I shouldn't *force* my data collection scripts to
alter when they think they got their datapoints... they happen
when they happen.)

later
steve
- - -
systems guy
wiscnet.net

--
* To unsubscribe from the rrd-users mailing list, send a message with the
  subject: unsubscribe to rrd-users-request at list.ee.ethz.ch