[rrd-users] Re: Preventing counter wraps

[Mike Wright]

At 12:00 AM 7/5/00 +0200, alex at ergens.op.Het.Net wrote:

>Originally you asked a way around these false wraps.  This should be
>done in the front end (perhaps with that "U", perhaps some other way).

I don't know if this has been discussed before or if it is already on the
TODO list, but I have been thinking that rrdtool itself should use the
sysUpTime of the device to determine the exact time between samples. 

ie: the frontend would query the interface in/out counters AND sysUpTime
(in a single SNMP query) then feed this data into rrdtool. RRDTool could
use the previous sysUpTime and counter values from the RRD to calculate the
bytes/second and store this in the RRA. 

This has the advantage that rrdtool could automatically detect reboots and
compensate for it. If RRDTool detected a reboot it could log the
bytes/second since reboot based on the value of the counters divided by
sysUptime to get the bytes/second since the reboot.

Also, sysUpTime gives us a 10ms sample resolution it would avoid the slight
jitter in the 1 second sampling we use at the moment (especially on high
latency/unreliable links).

Of course this wouldn't be any use for non SNMP devices which don't have a
sysUpTime and COUNTER type data sources, so it would have to be an optional
feature...

Cheers,

Mike Wright



--
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/rrd-users
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi